Datasets:
JevBench v1.6.0 results
Browse files- README.md +99 -0
- data/leaderboard.csv +95 -0
- data/leaderboard.json +1 -0
README.md
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
pretty_name: JevBench results
|
| 4 |
+
tags:
|
| 5 |
+
- leaderboard
|
| 6 |
+
- benchmark
|
| 7 |
+
- jev
|
| 8 |
+
- jev-class
|
| 9 |
+
- jev-alternatives
|
| 10 |
+
- decision-models
|
| 11 |
+
- model-evaluation
|
| 12 |
+
- calibration
|
| 13 |
+
- open-weights
|
| 14 |
+
task_categories:
|
| 15 |
+
- text-classification
|
| 16 |
+
size_categories:
|
| 17 |
+
- n<1K
|
| 18 |
+
configs:
|
| 19 |
+
- config_name: leaderboard
|
| 20 |
+
data_files:
|
| 21 |
+
- split: train
|
| 22 |
+
path: data/leaderboard.csv
|
| 23 |
+
---
|
| 24 |
+
|
| 25 |
+
# JevBench results
|
| 26 |
+
|
| 27 |
+
## Summary
|
| 28 |
+
|
| 29 |
+
System-level aggregate results of **JevBench**, Benchmark Heaven's open
|
| 30 |
+
benchmark for Jev-class decision models. Each row is one evaluated system with
|
| 31 |
+
its four axis scores (Intelligence, Calibration, Speed, Cost), the headline
|
| 32 |
+
**JevBench Capability Score** (arithmetic mean of Intelligence and Calibration,
|
| 33 |
+
ranked only within the official cost and median-latency caps) and the
|
| 34 |
+
secondary Composite score.
|
| 35 |
+
Data: JevBench v1.6.0, generated 2026-10-05.
|
| 36 |
+
|
| 37 |
+
`data/leaderboard.json` carries the full public feed as published at
|
| 38 |
+
`https://benchmarkheaven.com/api/jevbench/latest`.
|
| 39 |
+
|
| 40 |
+
## What is measured
|
| 41 |
+
|
| 42 |
+
- **Intelligence** — chance-corrected accuracy on the frozen public and sealed tiers.
|
| 43 |
+
- **Calibration** — distribution quality of the typed answers a system produces.
|
| 44 |
+
- **Speed** — median (p50) latency on a logarithmic scale; self-hosted runs carry a published adjustment.
|
| 45 |
+
- **Cost** — USD per 1,000 complete decisions; `price_kind` labels the evidence basis (`estimate` is an estimate, not a measured charge).
|
| 46 |
+
- **Capability score / rank** — the headline; rank is present only for systems inside the official caps.
|
| 47 |
+
- **Composite score / rank** — secondary ranking, Intelligence/Calibration/Speed/Cost weighted 25% each.
|
| 48 |
+
|
| 49 |
+
## Columns
|
| 50 |
+
|
| 51 |
+
| Column | Description |
|
| 52 |
+
| --- | --- |
|
| 53 |
+
| `revision` | JevBench release this row belongs to. |
|
| 54 |
+
| `key` | Stable machine key of the system. |
|
| 55 |
+
| `name` | Published system name. |
|
| 56 |
+
| `source_url` | Published source or weights link. |
|
| 57 |
+
| `listing` | `ranked` (in the ranking) or `listed` (shown, not ranked). |
|
| 58 |
+
| `capability_rank` | Capability rank; empty when the system is outside the official caps. |
|
| 59 |
+
| `capability_score` | Mean of Intelligence and Calibration (0–100). |
|
| 60 |
+
| `capability_eligible` | `true` when inside both official caps. |
|
| 61 |
+
| `composite_rank` | Composite rank (empty when not ranked). |
|
| 62 |
+
| `composite_score` | Composite score (0–100). |
|
| 63 |
+
| `intelligence`, `calibration`, `speed`, `cost_axis` | The four axis scores (0–100). |
|
| 64 |
+
| `usd_per_1000_decisions` | Modeled cost per 1,000 complete decisions. |
|
| 65 |
+
| `price_kind` | Evidence basis for the cost, e.g. `estimate`. |
|
| 66 |
+
| `p50_latency_s_raw` | Measured raw median latency in seconds (before any published adjustment). |
|
| 67 |
+
|
| 68 |
+
## How to cite
|
| 69 |
+
|
| 70 |
+
```bibtex
|
| 71 |
+
@misc{benchmarkheaven2026jevbench,
|
| 72 |
+
title = {JevBench: A Benchmark for Jev-class Decision Models},
|
| 73 |
+
author = {{Benchmark Heaven}},
|
| 74 |
+
year = {2026},
|
| 75 |
+
howpublished = {\url{https://benchmarkheaven.com/jev-models}},
|
| 76 |
+
note = {JevBench v1.6.0; regenerated at every JevBench release}
|
| 77 |
+
}
|
| 78 |
+
```
|
| 79 |
+
|
| 80 |
+
## Links
|
| 81 |
+
|
| 82 |
+
- [Live JevBench board](https://benchmarkheaven.com/jev-models)
|
| 83 |
+
- [Jev alternatives, compared](https://benchmarkheaven.com/jev-models/alternatives)
|
| 84 |
+
- [How to choose a Jev-class model](https://benchmarkheaven.com/jev-models/how-to-choose)
|
| 85 |
+
- [Is Jev open source?](https://benchmarkheaven.com/jev-models/open-source-jev)
|
| 86 |
+
- [Public results API](https://benchmarkheaven.com/api/jevbench/latest)
|
| 87 |
+
- [Submit a system](https://benchmarkheaven.com/submit)
|
| 88 |
+
- [Method repository](https://github.com/fstandhartinger/jevbench)
|
| 89 |
+
|
| 90 |
+
## Update cadence
|
| 91 |
+
|
| 92 |
+
Regenerated at every JevBench release.
|
| 93 |
+
|
| 94 |
+
## Limitations
|
| 95 |
+
|
| 96 |
+
Aggregates only: no item-level data (no questions, item ids or predictions).
|
| 97 |
+
Sealed items are never published. Cost values are modeled per the stated basis
|
| 98 |
+
and are not invoices; latency adjustments for self-hosted runs are published
|
| 99 |
+
assumptions, not measurements.
|
data/leaderboard.csv
ADDED
|
@@ -0,0 +1,95 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
revision,key,name,source_url,listing,capability_rank,capability_score,capability_eligible,composite_rank,composite_score,intelligence,calibration,speed,cost_axis,usd_per_1000_decisions,price_kind,p50_latency_s_raw
|
| 2 |
+
v1.6.0,quyet-1-0-large,"Quyet-1.0-Large (Chinh Nguyen, Gemma-4-31B decoder, option-letter logits)",https://huggingface.co/chinhnc/Quyet-1.0-Large,ranked,1,81.68880454670384,true,1,71.38706329141472,73.40135940607429,89.97624968733336,86.87440406834797,50.546097335562465,0.04451059729064039,estimate,0.11577823059633374
|
| 3 |
+
v1.6.0,jev-1.13.0,Jev 1.13.0 (TypeSafe AI),https://docs.typesafe.ai,ranked,3,76.48401425422394,true,2,71.08703004565037,62.40124449091305,90.56678401753481,91.49989299046987,54.72500234484654,0.032297327586206896,estimate,0.23910215869545937
|
| 4 |
+
v1.6.0,wity-1,"wity-1 (Wity, reasoning auto)",https://wity.alphanimble.com/,ranked,,79.76173748610775,false,3,70.91538581175126,71.55290364428411,87.97057132793138,71.79000307789984,58.368182152360596,0.02441896551724138,estimate,1.5746641159057617
|
| 5 |
+
v1.6.0,torchcast-decision-12b,"torchcast-decision-12b (Torchcast AI, Gemma-4-12B fine-tune, option-letter logprob readout)",https://huggingface.co/torchcast-ai/torchcast-decision-12b,ranked,4,71.73894611323924,true,4,69.9087143535159,60.54320091058725,82.93469131589123,91.74573681567192,56.36034169226667,0.02848756157635468,estimate,0.037151537369936705
|
| 6 |
+
v1.6.0,deck31b,"deck-31B (krishna765, frozen Gemma-4-31B-it, TorchAO FP8 dynamic)",https://github.com/krishna-gogineni-765/deck31b,ranked,2,77.5907429125261,true,5,69.33752640882658,73.02512554551848,82.15636027953371,86.69970536196755,49.89322575247467,0.04679784,estimate,0.12746820924803615
|
| 7 |
+
v1.6.0,winnow-12b,Winnow-12B Q8,https://huggingface.co/EldanRing/Winnow-12B,ranked,6,71.24586176250037,true,6,68.88542163459404,59.53366229166759,82.95806123333315,86.69753831262085,56.556023424877196,0.028062900246305422,estimate,0.1168902744539082
|
| 8 |
+
v1.6.0,cygnet,"Cygnet (blockbrain, frozen Gemma-4-12B-it)",https://github.com/blockbrain-ai/cygnet-recipe,ranked,7,70.90341055515565,true,7,68.55012424151383,54.83928122751086,86.96753988280044,91.77283664982562,56.42912569481113,0.028337561576354687,estimate,0.03681807057000697
|
| 9 |
+
v1.6.0,jev-omni,"Jev-Omni (akhilaaa3, Gemma-4-12B merged)",https://huggingface.co/akhilaaa3/Jev-Omni,ranked,5,71.2873662785918,true,8,67.74758858028281,55.530225828826175,87.04450672835743,85.43173201875288,56.05113155939739,0.02917173645320197,estimate,0.15469176834449172
|
| 10 |
+
v1.6.0,sage-1.3.0,Sage 1.3.0 (Levanto Labs),https://levanto.ai,ranked,,77.08747420427488,false,9,50.07295310117218,64.29339171700846,89.8815566915413,93.61638181239104,43.46555597245558,0.07664458333333334,estimate,0.15245595201849937
|
| 11 |
+
v1.6.0,plumb-4b,"Plumb-4B (crh225, JevK5 v0.2 + LoRA)",https://huggingface.co/crh225/plumb-4b,ranked,8,65.42184422831923,true,10,48.28224126395837,42.9626698747687,87.88101858186977,93.83356539601934,63.07359395281681,0.01701689039408867,estimate,0.019995882059447467
|
| 12 |
+
v1.6.0,spark-s1-4b-v6,"spark-s1-4b-v6 (Open Spark Jev, abhishek085)",https://github.com/abhishek085/open-spark-jev,ranked,30,54.016043277778124,true,11,44.90217636528639,43.2364603010071,64.79562625454915,86.95618624402076,60.422063196854964,0.020857610837438423,estimate,0.1286082404985791
|
| 13 |
+
v1.6.0,swanone,"swanOne (blockbrain, Qwen3.8-Flash-Next NVFP4)",https://github.com/blockbrain-ai/swanone-recipe,ranked,,72.36627654122822,false,12,43.78640157551217,56.15104294255194,88.58151013990448,82.64878475169937,42.15074720483203,0.08478293103448274,estimate,0.28074855310842395
|
| 14 |
+
v1.6.0,quyet-1-0-medium,"Quyet-1.0-Medium (Chinh Nguyen, Qwen3.5-4B decoder, option-letter logits)",https://huggingface.co/chinhnc/Quyet-1.0-Medium,ranked,21,59.24679183153445,true,13,43.58925238992216,41.73350604436517,76.76007761870373,89.46638461924755,63.43457194636937,0.01655189039408867,estimate,0.07741166599589633
|
| 15 |
+
v1.6.0,lev,"lev (Interfaze AI, Qwen3.5-4B + LoRA r32, label-token readout + candidate-path head, per-bucket temperatures)",https://huggingface.co/interfaze-ai/lev,ranked,18,61.70420936973613,true,14,42.177760514319104,41.084702071234844,82.32371666823741,88.00251499520694,61.79891447724963,0.01876590517241379,estimate,0.10230199046782218
|
| 16 |
+
v1.6.0,ninfer-qwen3.8-flash-next,NInfer Qwen3.8-Flash-Next mixed,https://github.com/igorls/ninfer,ranked,,69.67631278082561,false,15,41.96785076957165,48.980301026876646,90.37232453477458,89.33173525631216,42.53362878447125,0.08232764778325123,estimate,0.08002880541607738
|
| 17 |
+
v1.6.0,jev-local,jev-local (Qwen3.5-9B),https://github.com/us/jev-local,ranked,23,57.458381676033426,true,16,41.328564334669615,41.85041350559534,73.06634984647151,75.55955847735143,58.85554343293177,0.023522413793103446,estimate,0.3653302099555731
|
| 18 |
+
v1.6.0,decider-4b-v2,decider-4b v2 (Mapika),https://github.com/Mapika/decider,ranked,9,64.88166453375648,true,17,41.240894350857424,40.12597268917964,89.63735637833331,91.78549329698745,64.54376873042901,0.015201077586206898,estimate,0.024963717442005873
|
| 19 |
+
v1.6.0,jevk5-v0.3-4b,JevK5 v0.3 (4B),https://huggingface.co/alibiserikbay/JevK5,ranked,10,64.20716028123837,true,18,37.36493548751909,38.53201254374527,89.88230801873149,93.9472917853559,63.07359395281681,0.01701689039408867,estimate,0.018816601019352674
|
| 20 |
+
v1.6.0,metask-jev-4b,metask-jev-4b,https://github.com/metask-ai/metask-jev,ranked,17,61.79110407115073,true,19,37.34931073616377,39.166454612935595,84.41575352936586,90.78239008886905,57.7311327584882,0.025642610837438424,estimate,0.05397228497895412
|
| 21 |
+
v1.6.0,deck-4b-v1-0,"deck-4B v1.0 (krishna765, JevK5 + LoRA, FP8 weight-only)",https://github.com/krishna-gogineni-765/deck,ranked,12,63.60791322392738,true,20,37.29399798821111,38.62465958342338,88.59116686443139,89.93848373337222,63.66277849512568,0.0162645,estimate,0.07620866177603602
|
| 22 |
+
v1.6.0,clef-flash,"Clef-Flash (Cloudflare, Qwen3.5-9B post-train with a joint schema head, multimodal, measured on text)",https://huggingface.co/Cloudflare/clef-flash,ranked,11,63.824241207952895,true,21,36.635050479655426,42.18368888957246,85.46479352633332,88.4044196893156,46.79963593205177,0.05933971674876847,estimate,0.10283017950132489
|
| 23 |
+
v1.6.0,qwen35-9b-jev-data-mix-v2,Qwen3.5-9B Jev-like data-mix v2,https://huggingface.co/jsaurabh/qwen3.5-9b-jev-data-mix-v2,ranked,,60.86540021228136,false,22,33.58966697347588,41.94184606895763,79.78895435560509,85.40793837616167,45.690218065856605,0.06461391625615764,estimate,0.16828587953932583
|
| 24 |
+
v1.6.0,jevk5-v02,JevK5 v0.2.0,https://github.com/allebee/jevk5,ranked,13,62.824112319658205,true,23,32.09901620353128,36.261123811123895,89.38710082819252,91.56570599843724,63.07359395281681,0.01701689039408867,estimate,0.04039452798315324
|
| 25 |
+
v1.6.0,jevone,JevOne,https://huggingface.co/juspay/jev-one,ranked,,68.8604478570867,false,24,31.14510331640707,45.955077428728934,91.76581828544445,89.83885080776159,39.84212731591276,0.10121908866995075,estimate,0.06179859535768628
|
| 26 |
+
v1.6.0,hopper,Hopper,https://huggingface.co/HopitAI/hopper,ranked,14,62.40936838189898,true,25,28.852652318563656,34.73970870443837,90.07902805935959,91.11434937226889,62.25627699896333,0.018118577586206896,estimate,0.0564777513500303
|
| 27 |
+
v1.6.0,imajev-4b-rtx5090-a2,Imajev-4B (RTX 5090),https://github.com/fstandhartinger/jevbench/issues/110,ranked,15,61.89494846465412,true,26,28.655751132526984,34.62714018309005,89.1627567462182,90.51878000442085,63.26458220826236,0.016769261083743846,estimate,0.07120202149963006
|
| 28 |
+
v1.6.0,malkuth-4b,"Malkuth-4B (newfull5, Kev post-train)",https://github.com/newfull5/malkuth,ranked,19,61.139059860562405,true,27,26.197476126697616,33.840703805987175,88.43741591513763,89.51571180654442,55.82433433950372,0.02968398399014778,estimate,0.07427763848681934
|
| 29 |
+
v1.6.0,jqv,jqv (Qwen3-32B zero-shot),https://github.com/Octalab-Inc/jqv,ranked,16,61.8647697220168,true,28,25.391439220985372,33.85815624201783,89.87138320201578,83.1032423977654,51.172592607797476,0.042420935960591134,estimate,0.09740161104127765
|
| 30 |
+
v1.6.0,decider-35b-a3b,decider-35b-a3b (Mapika),https://huggingface.co/Mapika/decider-35b-a3b,ranked,,66.1715118587882,false,29,24.772887446577812,48.72471958402083,83.61830413355554,91.27357492980725,34.38579876859529,0.1538650708128079,estimate,0.04968994203954935
|
| 31 |
+
v1.6.0,manchego-v21-v15,Manchego v2.1,https://github.com/nschlaepfer/manchego-serve,ranked,22,58.27541943544253,true,30,24.487333472500133,32.54303915979197,84.0077997110931,90.7347510499745,64.32996474661243,0.015452586206896551,estimate,0.05371422949974658
|
| 32 |
+
v1.6.0,decision-4b-v12,"Decision 4B v1.2 (FlyMyJev, Qwen3.5-4B + LoRA)",https://benchmarkheaven.com/jev-models/decision-4b-v12,ranked,20,60.41399746452394,true,31,24.463556699266523,32.40310972936693,88.42488519968094,93.92936299294371,63.07359395281681,0.01701689039408867,estimate,0.019149313447996974
|
| 33 |
+
v1.6.0,raw-qwen3-4b-instruct-2507,Raw Qwen3 4B Instruct 2507 direct logits,https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507,ranked,53,34.56646466717713,true,32,21.792436038282993,33.95117742100547,35.18175191334879,91.16296741568695,63.36068112101963,0.01664602832512315,estimate,0.04603758594021201
|
| 34 |
+
v1.6.0,nimble-9b,Bespoke Nimble 9B (Bespoke Labs),https://github.com/bespokelabsai/nimble,ranked,,56.514998578191054,false,33,21.082616704957896,43.14237010986392,69.88762704651819,86.11327010751721,36.75037423027614,0.1283278325123153,estimate,0.14493575389496982
|
| 35 |
+
v1.6.0,reflex-4b,reflex 4B (kshetrajna12),https://github.com/kshetrajna12/reflex,ranked,,58.95740703790636,false,34,20.637311182702398,30.935558296483705,86.97925577932901,68.7900017304639,63.142712976263866,0.016926853448275863,estimate,1.3577107690507546
|
| 36 |
+
v1.6.0,typecastlm,"typecastlm (Mikhail Gribov, Qwen3.5-4B computed head)",https://github.com/mihail-gribov/typecastlm,ranked,26,56.544258540063545,true,35,19.72256430004839,29.78900656204624,83.29951051808085,92.76354061232337,63.95900966831593,0.01589887315270936,estimate,0.02951649052556604
|
| 37 |
+
v1.6.0,decision-4b-v11,"Decision 4B v1.1 (FlyMyJev, Qwen3.5-4B + LoRA)",https://benchmarkheaven.com/jev-models/decision-4b-v11,ranked,24,57.18675493386165,true,36,18.718356274892123,29.134168280172865,85.23934158755043,93.95641941218838,63.07359395281681,0.01701689039408867,estimate,0.019026647554710507
|
| 38 |
+
v1.6.0,autojev-27b,"AutoJev-27B (denis-pplx, Qwen3.8-27B)",https://huggingface.co/denis-pplx/autojev-27b,ranked,,71.61983491755967,false,37,18.463427917707932,55.62422706919757,87.61544276592178,88.60757446254297,29.365937020749882,0.22618758620689655,estimate,0.09737715951632708
|
| 39 |
+
v1.6.0,eikos-27b,"Eikos-27B (caiovicentino1, Qwen3.8-27B)",https://huggingface.co/caiovicentino1/eikos-27b,ranked,,77.18992599407068,false,38,18.077889022432696,64.40270761245412,89.97714437568723,88.08630815059152,28.689752882469676,0.23823646551724137,estimate,0.11261826660484076
|
| 40 |
+
v1.6.0,ninfer-qwen3.8-27b,NInfer Qwen3.8-27B NVFP4,https://github.com/igorls/ninfer,ranked,,67.99284998901135,false,39,17.71600496541158,51.800126951055375,84.18557302696732,90.47734239273063,29.118887844204664,0.2305174137931034,estimate,0.05501334450673312
|
| 41 |
+
v1.6.0,openjev-thinking,"OpenJev (thinking, BF16)",https://github.com/razorback16/openjev,ranked,,75.87983429229429,false,40,17.265218966202397,70.34621384425424,81.41345474033434,72.55270827201397,28.51480924396573,0.24145693965517237,estimate,0.8569832294015214
|
| 42 |
+
v1.6.0,open-jev-zefan-9b,Open-Jev 9B (Zefan Cai),https://github.com/Zefan-Cai/Open-Jev,ranked,,61.84549459243814,false,41,17.045611746249723,43.88591848753404,79.80507069734225,74.39486349899005,33.05490478272432,0.17041317733990147,estimate,0.6175056385109201
|
| 43 |
+
v1.6.0,standardone-8b,Standard One 8B (Standard Thinking),https://huggingface.co/StandardThinking/StandardOne-8B,ranked,,58.40967348707115,false,42,16.917108299009165,32.80426604507879,84.01508092906352,92.76442849252426,43.283250331904235,0.07772456896551724,estimate,0.03173699742183089
|
| 44 |
+
v1.6.0,clef,"Clef (Cloudflare, Qwen3.8-27B post-train with a joint schema head, multimodal, measured on text)",https://huggingface.co/Cloudflare/clef,ranked,,75.32934491938597,false,43,16.753692861922005,59.08165501447566,91.57703482429629,83.3998209667487,28.102157220114762,0.24922681034482758,estimate,0.21318734902888536
|
| 45 |
+
v1.6.0,jev-qwen3.5-9b-base-nvfp4,JEV Qwen3.5-9B Base NVFP4,https://huggingface.co/WIlfLin/JEV-Qwen3.5-9B-Base-NVFP4,ranked,25,57.14200691217819,true,44,15.97555620862577,29.251131763478327,85.03288206087805,93.62779491136789,47.592235403520306,0.05583743842364531,estimate,0.02267279999796301
|
| 46 |
+
v1.6.0,semif-qwen3.5-4b,"SemIf, formerly OpenJev (Qwen3.5-4B, TheoLeeCJ)",https://github.com/TheoLeeCJ/openjev,ranked,27,55.35036926555066,true,45,15.517179498801362,27.10113669038628,83.59960184071504,90.76560119421038,63.07359395281681,0.01701689039408867,estimate,0.05204228049842641
|
| 47 |
+
v1.6.0,bev-bonsai-27b,Bev / Bonsai 27B,https://huggingface.co/Reza2kn/Bev,ranked,,66.13873250860377,false,46,11.7393704396132,43.77513601921869,88.50232899798883,72.98222461463935,28.231834531826323,0.2467585344827586,estimate,0.9861058448441327
|
| 48 |
+
v1.6.0,litjev,LitJev (Qwen3.8-27B),https://github.com/zhengxuyu/litjev,ranked,,66.60329337630368,false,47,11.458886170296882,43.28549422942267,89.9210925231847,68.17671225113584,28.35809912639418,0.2443787068965517,estimate,1.4730074456892908
|
| 49 |
+
v1.6.0,localjev-qwen3.5-4b,local-jev Qwen3.5-4B,https://github.com/amithgc/local-jev,ranked,28,54.99411342282565,true,48,11.438161297684031,24.145363218466134,85.84286362718517,86.10068849774444,59.269627658320445,0.022786576354679802,estimate,0.1226705745793879
|
| 50 |
+
v1.6.0,system-one-sg,"system-one (Qwen3-8B, Sean Goedecke)",https://github.com/sgoedecke/system-one,ranked,,29.718120999285617,false,49,11.027723483913586,29.271508779158296,30.16473321941294,90.8867694864598,44.96855803039237,0.06829381896551724,estimate,0.04996280049090274
|
| 51 |
+
v1.6.0,kev-8b,kev 8B (research preview),https://github.com/jaredpalmer/kev,ranked,,50.0091761607623,false,50,10.62283914209564,29.26217941280341,70.7561729087212,88.97143473610706,40.42054386073457,0.09682374384236453,estimate,0.07448393944650888
|
| 52 |
+
v1.6.0,decision-2b,"Decision 2B (FlyMy.AI, v59)",https://huggingface.co/flymy-ai/decision-2b-preview,ranked,29,54.82290333495299,true,51,10.289708493744515,22.875473256337628,86.77033341356835,90.39462801231525,66.37275756963365,0.013210160098522166,estimate,0.07514283142518252
|
| 53 |
+
v1.6.0,opensourcejev-qwen35-4b-q4km,"OpenSourceJev (Qwen3.5-4B Q4_K_M, native llama.cpp)",https://github.com/sabeel111/OpenSourceJev,ranked,33,48.571422117188824,true,52,10.263493713193125,23.217995107845883,73.92484912653177,77.06362388237805,69.13657649738097,0.0106851539408867,estimate,0.29475566232576966
|
| 54 |
+
v1.6.0,open-alternative-jev,"open-alternative-jev (Qwen3.5-4B, IkerMoel)",https://github.com/ikermoel/open-alternative-jev,ranked,32,49.14497164755345,true,53,9.435692571873393,22.34610151162883,75.94384178347806,91.30216983336405,63.27859776193043,0.0167512315270936,estimate,0.04333395400317386
|
| 55 |
+
v1.6.0,raw-qwen3-8b,Raw Qwen3 8B direct logits,https://huggingface.co/Qwen/Qwen3-8B,ranked,,30.13533882926388,false,54,9.30125972789427,26.54912920590792,33.721548452619835,88.65282990878164,45.53515592313027,0.0653875104679803,estimate,0.07753071398474276
|
| 56 |
+
v1.6.0,decider-2b,decider-2b (Mapika),https://huggingface.co/Mapika/decider-2b,ranked,37,44.58947913615145,true,55,7.6704332123718,20.654576369710338,68.52438190259257,94.55945724666857,64.91766177740826,0.014771046798029557,estimate,0.014745104941539466
|
| 57 |
+
v1.6.0,nemotron-diffusion-8b,"Nemotron Diffusion 8B (pst2154, optimized vLLM)",https://github.com/pst2154/Nemotron_Jev,ranked,35,47.46294541124625,true,56,7.29748476516348,20.38848344571935,74.53740737677316,93.8305724848098,55.49170925809556,0.03045157019704434,estimate,0.02195067796856165
|
| 58 |
+
v1.6.0,kev-4b,kev 4B (research preview),https://github.com/jaredpalmer/kev,ranked,39,43.98178354242468,true,57,6.611791819903415,19.510707720813777,68.45285936403559,90.46585183268016,65.78036497095258,0.013824661330049262,estimate,0.07199650770053267
|
| 59 |
+
v1.6.0,malkuth-2b,"Malkuth-2B (newfull5, Kev post-train)",https://github.com/newfull5/malkuth,ranked,34,47.633125198824516,true,58,3.992169201455145,15.917395195670291,79.34885520197874,92.03462875001455,65.5661191189412,0.014053873152709357,estimate,0.04305429550004192
|
| 60 |
+
v1.6.0,qwen3-reranker-4b,Qwen3-Reranker-4B,https://huggingface.co/Qwen/Qwen3-Reranker-4B,ranked,36,44.7457252992122,true,59,3.715006798009075,16.337944014530347,73.15350658389404,82.14971164581814,48.39770052432463,0.05249000923645321,estimate,0.20409131213091314
|
| 61 |
+
v1.6.0,zerank-2,ZeroEntropy zerank-2,https://huggingface.co/zeroentropy/zerank-2-reranker,ranked,31,49.39850940998648,true,60,3.349795360318651,15.596611891228248,83.20040692874471,82.89783918855927,48.39770052432463,0.05249000923645321,estimate,0.17457342706620693
|
| 62 |
+
v1.6.0,open-jev-zefan-2b,Open-Jev 2B (Zefan Cai),https://github.com/Zefan-Cai/Open-Jev,ranked,,47.71137138509946,false,61,3.2383990318693976,21.804441188912403,73.61830158128652,77.13731150142542,33.05490478272432,0.17041317733990147,estimate,0.43650909198913723
|
| 63 |
+
v1.6.0,open-jev-zefan-27b-v1.1,"Open-Jev 27B v1.1 (Zefan Cai, LoRA + decision head on Qwen3.8-27B)",https://huggingface.co/ZefanCai/Open-Jev-27B-v1.1,ranked,,67.96306879751505,false,62,2.8808671915746493,52.45777412397041,83.46836347105969,72.50871441156775,14.357426070787312,0.7157353448275862,estimate,0.7902996460907161
|
| 64 |
+
v1.6.0,raw-phi-4-mini,Raw Phi-4 mini direct logits,https://huggingface.co/microsoft/Phi-4-mini-instruct,ranked,52,35.33785192837895,true,63,2.4794266829120515,13.660237978082595,57.015465878675315,91.89255362835293,53.22051180195924,0.03625068965517241,estimate,0.04178677382878959
|
| 65 |
+
v1.6.0,simplejev-qwen3.5-0.8b,"SimpleJev (Qwen3.5-0.8B, CPU)",https://github.com/featherless-ai/simple-jev,ranked,,31.59915405528791,false,64,2.112084174105171,13.000426205687345,50.197881904888476,58.929227255635794,70.31175519898045,0.009763559113300492,estimate,4.323511580238119
|
| 66 |
+
v1.6.0,smalljev,smalljev semantic-v9,https://github.com/isHeSatoshi/smalljev,ranked,43,42.95073007048671,true,65,0.7719455446134506,8.670533657732655,77.23092648324078,90.21131723708834,60.77469204714053,0.020300665024630543,estimate,0.06657194206491113
|
| 67 |
+
v1.6.0,kev-0.6b,kev 0.6B (research preview),https://github.com/jaredpalmer/kev,ranked,51,36.93267237588606,true,66,0.5431213026023877,7.598789169415934,66.26655558235618,91.51497362122684,80.09400261254245,0.004608220443349753,estimate,0.05674442206509411
|
| 68 |
+
v1.6.0,opendecision,OpenDecision (ModernBERT-large zero-shot),https://github.com/deepanwadhwa/OpenDecision,ranked,48,37.99564488451158,true,67,0.508795165212109,7.425225288997898,68.56606448002526,87.12856375997052,79.10704862374335,0.004970862068965518,estimate,0.07184981391765177
|
| 69 |
+
v1.6.0,mghafiri-qwen3.5-0.8b-decision-model,Qwen3.5-0.8B Decision Model (Mourad Ghafiri),https://huggingface.co/mghafiri/qwen3.5-0.8B-decision-model,ranked,,42.62692401157305,false,68,0.47247471382875306,7.297270616183145,77.95657740696296,55.90747410573378,79.5180630553356,0.00481649630541872,estimate,5.551045611966401
|
| 70 |
+
v1.6.0,deem-0.8-v1,Deem 0.8B v1,https://huggingface.co/LibertAIDAI/deem-0.8-v1,ranked,60,18.07951706793502,true,69,0.32342918922131697,6.531204174677637,29.6278299611924,85.44505463939521,80.28784730320886,0.004540166256157636,estimate,0.15738983592018485
|
| 71 |
+
v1.6.0,decision-fast,"Decision Fast (FlyMy.AI, v53a)",https://huggingface.co/flymy-ai/decision-fast-preview,ranked,45,40.59403173266273,true,70,0.2981489585308357,6.110691078442272,75.07737238688318,91.472185793536,80.08651867745036,0.004610868226600985,estimate,0.05768406600691378
|
| 72 |
+
v1.6.0,quyet-1-0-small-en,"Quyet-1.0-Small-EN (Chinh Nguyen, ModernBERT-base encoder with fixed heads)",https://huggingface.co/chinhnc/Quyet-1.0-Small-EN,ranked,47,39.53807694084099,true,71,0.29239342204088814,6.054813340422714,73.02134054125926,94.90893356488051,89.11296503596958,0.0023062222906403943,estimate,0.0145730719959829
|
| 73 |
+
v1.6.0,fastino-gliner-2-5-decide,Fastino GLiNER-2.5-Decide (hosted API),https://fastino.ai/,ranked,49,37.63882868110011,true,72,0.28381932562808126,6.439988814479404,68.83766854772081,84.7052606179839,46.59842351525712,0.06026325,estimate,0.4409300498664379
|
| 74 |
+
v1.6.0,laya-typed-decisions,Laya typed-decisions,https://huggingface.co/convaiinnovations/laya/tree/1c5edc17a7acd8701df6fc341c0d179f1c62c982/typed-decisions,ranked,,44.22061775531203,false,73,0.21666109526429145,5.471270906587037,82.96996460403702,70.81965680870755,82.52942826196218,0.003822543103448276,estimate,0.8638670064901817
|
| 75 |
+
v1.6.0,openjev-verdict-1.4,openJev Verdict 1.4,https://huggingface.co/heman10x/rlcd-modernbert-151m,ranked,46,40.36740412437111,true,74,0.17222636557331997,5.035901218168153,75.69890703057408,81.49274954261848,86.62313527343974,0.002791871921182266,estimate,0.2926154159940779
|
| 76 |
+
v1.6.0,raw-qwen3-0.6b,Raw Qwen3 0.6B direct logits,https://huggingface.co/Qwen/Qwen3-0.6B,ranked,63,8.247746904337834,true,75,0.1673536172919218,5.554864880982027,10.94062892769364,93.52703208725202,77.58073177551512,0.005588676108374384,estimate,0.02898204093798995
|
| 77 |
+
v1.6.0,lev-350m,"lev-350m (Franck Verrot, LFM2.5-350M)",https://github.com/franckverrot/lev,ranked,41,43.34590358011229,true,76,0.15377137469809216,4.827562367090417,81.86424479313416,93.98230372494703,80.03797760430007,0.00462807881773399,estimate,0.022326925001834752
|
| 78 |
+
v1.6.0,raw-qwen3-1.7b,Raw Qwen3 1.7B direct logits,https://huggingface.co/Qwen/Qwen3-1.7B,ranked,62,8.320306683538828,true,77,0.13671939226506033,5.124286554530066,11.516326812547591,93.23530725162705,68.54983190559568,0.011177352216748768,estimate,0.03198186121881008
|
| 79 |
+
v1.6.0,kev-0.5b,kev 0.5B,https://github.com/jaredpalmer/kev,ranked,54,32.90230956430239,true,78,0.131588623265787,4.598433868646836,61.20618525995795,92.39904218127765,80.09398520305423,0.004608226600985222,estimate,0.043983193347230554
|
| 80 |
+
v1.6.0,openjev-verdict,"openJev Verdict (heman10x, ModernBERT-base 151M)",https://github.com/Heman10x-NGU/openJev-verdict-2.0,ranked,58,26.63220276955003,true,79,0.09650709186186587,4.1496399816416485,49.11476555745841,79.40046783928,86.62313527343974,0.002791871921182266,estimate,0.2692721150815487
|
| 81 |
+
v1.6.0,quyet-1-0-small,"Quyet-1.0-Small (Chinh Nguyen, SEA-LION-ModernBERT-300M encoder with fixed heads)",https://huggingface.co/chinhnc/Quyet-1.0-Small,ranked,55,32.492363520877504,true,80,0.08963716950381444,4.019045994532796,60.96568104722222,94.91846454857637,79.56261245432609,0.004800055418719211,estimate,0.01451935400837101
|
| 82 |
+
v1.6.0,quyet-1-0-tiny,"Quyet-1.0-Tiny (Chinh Nguyen, mmBERT-small encoder (16 layers kept) with fixed heads)",https://huggingface.co/chinhnc/Quyet-1.0-Tiny,ranked,56,31.08052871088324,true,81,0.07064766318734213,3.6964771541368573,58.46458026762962,95.09610910731021,88.43590631389968,0.0024292364532019703,estimate,0.012666817499848548
|
| 83 |
+
v1.6.0,open-jev-deberta-v3-large,open-jev-deberta-v3-large (local CPU),https://github.com/kotoba-lang/typed-decisions,ranked,,35.39766066495959,false,82,0.03235814777127837,2.8293414417857914,67.96597988813339,67.62138193768119,77.5922354035203,0.005583743842364532,estimate,1.6890084934420884
|
| 84 |
+
v1.6.0,verdict-small,"verdict-small (Manavarya09, multilingual-e5-small 118M)",https://github.com/Manavarya09/verdict,ranked,50,37.63085686322157,true,83,0.01899384266854915,2.3417095945172277,72.92000413192591,89.48731594094681,100.0,0.000866152709359606,estimate,0.04910206887871027
|
| 85 |
+
v1.6.0,laya,"Laya (Convai Innovations, ModernBERT-large 421M)",https://huggingface.co/convaiinnovations/laya,ranked,,32.50714989618597,false,84,0.008560099345041717,1.7910341507793501,63.22326564159259,73.26302890816496,84.89402784895724,0.0031881034482758625,estimate,0.825308452360332
|
| 86 |
+
v1.6.0,mxbai-rerank-base-v2,Mixedbread mxbai-rerank-base-v2,https://huggingface.co/mixedbread-ai/mxbai-rerank-base-v2,ranked,40,43.87421595931999,true,85,0.002273929358844764,1.140707327109041,86.60772459153094,92.58758170313658,60.33589314238483,0.020996016009852216,estimate,0.031425865134224296
|
| 87 |
+
v1.6.0,laya-multilingual,Laya multilingual,https://huggingface.co/convaiinnovations/laya/tree/1c5edc17a7acd8701df6fc341c0d179f1c62c982/multilingual,ranked,61,16.971882566936873,true,86,3.7595668430455566e-05,0.2879184668737643,33.655846666999985,77.82044954853706,82.1551651086044,0.003933940886699508,estimate,0.3577279045130126
|
| 88 |
+
v1.6.0,certo,Certo v1 (AltSlate Labs),https://huggingface.co/altslate/certo-decision-model,ranked,42,43.17437163074946,true,87,1.3254531668229452e-05,0.20278658812856143,86.14595667337036,92.53493469610669,96.9130801359906,0.0012673522167487683,estimate,0.042036901926621795
|
| 89 |
+
v1.6.0,clm-8b,"CLM-8B (Contrastive-LM, clm-latest)",https://github.com/Contrastive-LM/CLM,ranked,59,20.086328406467768,true,88,3.5510969933114796e-06,0.13075619159747265,40.041900621338065,92.63548383949004,50.501662240594015,0.044662660714285714,estimate,0.02119073923677206
|
| 90 |
+
v1.6.0,bge-reranker-v2-m3,BAAI bge-reranker-v2-m3,https://huggingface.co/BAAI/bge-reranker-v2-m3,ranked,38,44.08279577608597,true,89,4.4076803151681546e-07,0.06512222845080358,88.10046932372114,93.86976236118387,59.325605115070715,0.022688885467980296,estimate,0.01707328436896205
|
| 91 |
+
v1.6.0,gte-reranker-modernbert-base,Alibaba GTE Reranker ModernBERT-base,https://huggingface.co/Alibaba-NLP/gte-reranker-modernbert-base,ranked,44,41.7278116825546,true,90,0.0,0.0,83.4556233651092,93.76120791185616,69.26965431492843,0.010576570197044334,estimate,0.024179050931707025
|
| 92 |
+
v1.6.0,mirror,Mirror,https://github.com/Bluusun/Decision-API,ranked,,25.806386196486883,false,91,0.0,0.0,51.61277239297377,73.47211485993975,89.27300805786774,0.0022780665024630543,estimate,0.6508294129744172
|
| 93 |
+
v1.6.0,open-jev-json-canvas-joshuasp,Open Jev JSON Canvas (JoshuaSP),https://github.com/JoshuaSP/open-jev,ranked,57,30.596109035139946,true,92,0.0,61.19221807027989,0.0,85.53332924416175,49.295219536461964,0.04899585591133005,estimate,0.14833511249162257
|
| 94 |
+
v1.6.0,wity-1-always,"wity-1 (Wity, reasoning always)",https://wity.alphanimble.com/,listed,,80.71241810169991,false,,70.96883700634847,70.45905034996007,90.96578585343975,71.20296182763504,58.368182152360596,0.02441896551724138,estimate,1.6885605081915855
|
| 95 |
+
v1.6.0,wity-1-off,"wity-1 (Wity, reasoning off)",https://wity.alphanimble.com/,listed,,57.913185425437945,false,,40.53344853152938,40.91725354389175,74.90911730698414,89.55912090143326,58.368182152360596,0.02441896551724138,estimate,0.28173011541366577
|
data/leaderboard.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"schema_version":1,"benchmark":"JevBench","revision":"v1.6.0","source":{"page":"/jev-models","frozen_page":"/jev-models/v1.6.0","artifact":"/api/jevbench/v1.6.0","artifact_sha256":"b8560f6e00c45d5dfdf254954d2176b015aad279c741873c0ed7345673354866"},"headline":"capability","composite_option":"A","weights":{"intelligence":25,"calibration":25,"speed":25,"cost":25},"capability_policy":{"definition":"Arithmetic mean of Intelligence and Calibration; ranked models within both official caps","reference_key":"jev-1.13.0@v1.5.7","cost_cap_usd_per_1000":0.06459465517241379,"median_latency_cap_s":1.2329566404223442,"factor":2},"systems":[{"key":"quyet-1-0-large","name":"Quyet-1.0-Large (Chinh Nguyen, Gemma-4-31B decoder, option-letter logits)","source_url":"https://huggingface.co/chinhnc/Quyet-1.0-Large","axes":{"intelligence":73.40135940607429,"calibration":89.97624968733336,"speed":86.87440406834797,"cost":50.546097335562465},"capability":{"score":81.68880454670384,"rank":1,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":71.38706329141472,"rank":1,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.04451059729064039,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): OpenRouter google/gemma-4-31b-it list price $0.09/M input, $0.34/M output (the exact base model google/gemma-4-31B-it; listed since Apr 2026), read 2026-10-04 x the package's own measured usage.input_tokens; one forward pass, nothing generated; run in process on 1x H100 80GB (Lium), quyet 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.11577823059633374,"p50_s_adjusted":0.3815564611926675,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jev-1.13.0","name":"Jev 1.13.0 (TypeSafe AI)","source_url":"https://docs.typesafe.ai","axes":{"intelligence":62.40124449091305,"calibration":90.56678401753481,"speed":91.49989299046987,"cost":54.72500234484654},"capability":{"score":76.48401425422394,"rank":3,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":71.08703004565037,"rank":2,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.032297327586206896,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): operator standard launch list price (interpretation I-1); no exact base-model floor applies"},"latency":{"p50_s_raw":0.23910215869545937,"p50_s_adjusted":0.23910215869545937,"adjustment":"none (API)"}},{"key":"wity-1","name":"wity-1 (Wity, reasoning auto)","source_url":"https://wity.alphanimble.com/","axes":{"intelligence":71.55290364428411,"calibration":87.97057132793138,"speed":71.79000307789984,"cost":58.368182152360596},"capability":{"score":79.76173748610775,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 2.6× Jev (v1.5 reference)"]},"composite_score":70.91538581175126,"rank":3,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.02441896551724138,"basis":"Wity's stated API tariff: USD 0.042 per 1M input tokens, output free; v1.5 cost per 1,000 decisions carried from the v1.5.8 build (measured 944,200 input tokens over 1,624 decisions, build f65ce826a455). API models with a known base are ranked at their own API price (1 Oct 2026 rule)."},"latency":{"p50_s_raw":1.5746641159057617,"p50_s_adjusted":1.5746641159057617,"adjustment":"none (API)"}},{"key":"torchcast-decision-12b","name":"torchcast-decision-12b (Torchcast AI, Gemma-4-12B fine-tune, option-letter logprob readout)","source_url":"https://huggingface.co/torchcast-ai/torchcast-decision-12b","axes":{"intelligence":60.54320091058725,"calibration":82.93469131589123,"speed":91.74573681567192,"cost":56.36034169226667},"capability":{"score":71.73894611323924,"rank":4,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":69.9087143535159,"rank":4,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.02848756157635468,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; no public tariff, customer asked for the base-model reference): OpenRouter google/gemma-3-12b-it list price $0.05/M input, $0.15/M output (gemma-4-12b-it is not listed on OpenRouter, read 2026-10-04; same reference as the Cygnet and Jev-Omni Gemma-4-12B rows) x the shim's own measured usage.prompt_tokens/completion_tokens; self-hosted on 1x H100 80GB (Lium), vLLM 0.30.0 BF16; estimated, not charged"},"latency":{"p50_s_raw":0.037151537369936705,"p50_s_adjusted":0.2243030747398734,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"deck31b","name":"deck-31B (krishna765, frozen Gemma-4-31B-it, TorchAO FP8 dynamic)","source_url":"https://github.com/krishna-gogineni-765/deck31b","axes":{"intelligence":73.02512554551848,"calibration":82.15636027953371,"speed":86.69970536196755,"cost":49.89322575247467},"capability":{"score":77.5907429125261,"rank":2,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":69.33752640882658,"rank":5,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.04679784,"basis":"ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): OpenRouter google/gemma-4-31b-it list price $0.09/M input, $0.34/M output (the exact frozen base model google/gemma-4-31B-it; same reference as Quyet-1.0-Large), read 2026-10-04 x the server's own measured usage.input_tokens; letter-mass readout, nothing generated; TorchAO FP8 dynamic on 1x GPU (Lium), deck31b 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.12746820924803615,"p50_s_adjusted":0.4049364184960723,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"winnow-12b","name":"Winnow-12B Q8","source_url":"https://huggingface.co/EldanRing/Winnow-12B","axes":{"intelligence":59.53366229166759,"calibration":82.95806123333315,"speed":86.69753831262085,"cost":56.556023424877196},"capability":{"score":71.24586176250037,"rank":6,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":68.88542163459404,"rank":6,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.028062900246305422,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.1168902744539082,"p50_s_adjusted":0.38378054890781643,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"cygnet","name":"Cygnet (blockbrain, frozen Gemma-4-12B-it)","source_url":"https://github.com/blockbrain-ai/cygnet-recipe","axes":{"intelligence":54.83928122751086,"calibration":86.96753988280044,"speed":91.77283664982562,"cost":56.42912569481113},"capability":{"score":70.90341055515565,"rank":7,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":68.55012424151383,"rank":7,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.028337561576354687,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.03681807057000697,"p50_s_adjusted":0.22363614114001393,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jev-omni","name":"Jev-Omni (akhilaaa3, Gemma-4-12B merged)","source_url":"https://huggingface.co/akhilaaa3/Jev-Omni","axes":{"intelligence":55.530225828826175,"calibration":87.04450672835743,"speed":85.43173201875288,"cost":56.05113155939739},"capability":{"score":71.2873662785918,"rank":5,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":67.74758858028281,"rank":8,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.02917173645320197,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.15469176834449172,"p50_s_adjusted":0.45938353668898346,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"sage-1.3.0","name":"Sage 1.3.0 (Levanto Labs)","source_url":"https://levanto.ai","axes":{"intelligence":64.29339171700846,"calibration":89.8815566915413,"speed":93.61638181239104,"cost":43.46555597245558},"capability":{"score":77.08747420427488,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.4× Jev (v1.5 reference)"]},"composite_score":50.07295310117218,"rank":9,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.07664458333333334,"basis":"Levanto list tariff, top tier: USD 0.05 per 1M input tokens, USD 10 per 1M output tokens (https://levanto.ai/pricing, read 5 Oct 2026)"},"latency":{"p50_s_raw":0.15245595201849937,"p50_s_adjusted":0.15245595201849937,"adjustment":"none (API)"}},{"key":"plumb-4b","name":"Plumb-4B (crh225, JevK5 v0.2 + LoRA)","source_url":"https://huggingface.co/crh225/plumb-4b","axes":{"intelligence":42.9626698747687,"calibration":87.88101858186977,"speed":93.83356539601934,"cost":63.07359395281681},"capability":{"score":65.42184422831923,"rank":8,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":48.28224126395837,"rank":10,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.019995882059447467,"p50_s_adjusted":0.18999176411889493,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"spark-s1-4b-v6","name":"spark-s1-4b-v6 (Open Spark Jev, abhishek085)","source_url":"https://github.com/abhishek085/open-spark-jev","axes":{"intelligence":43.2364603010071,"calibration":64.79562625454915,"speed":86.95618624402076,"cost":60.422063196854964},"capability":{"score":54.016043277778124,"rank":30,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":44.90217636528639,"rank":11,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.020857610837438423,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.1286082404985791,"p50_s_adjusted":0.40721648099715824,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"swanone","name":"swanOne (blockbrain, Qwen3.8-Flash-Next NVFP4)","source_url":"https://github.com/blockbrain-ai/swanone-recipe","axes":{"intelligence":56.15104294255194,"calibration":88.58151013990448,"speed":82.64878475169937,"cost":42.15074720483203},"capability":{"score":72.36627654122822,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.6× Jev (v1.5 reference)"]},"composite_score":43.78640157551217,"rank":12,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.08478293103448274,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.28074855310842395,"p50_s_adjusted":0.7114971062168479,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"quyet-1-0-medium","name":"Quyet-1.0-Medium (Chinh Nguyen, Qwen3.5-4B decoder, option-letter logits)","source_url":"https://huggingface.co/chinhnc/Quyet-1.0-Medium","axes":{"intelligence":41.73350604436517,"calibration":76.76007761870373,"speed":89.46638461924755,"cost":63.43457194636937},"capability":{"score":59.24679183153445,"rank":21,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":43.58925238992216,"rank":13,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01655189039408867,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): DeepInfra Qwen/Qwen3.5-4B list price $0.03/M input (the exact base weights; not on OpenRouter; same reference as reflex-4b, jobe and typecastlm), read 2026-09-24 x the package's own measured usage.input_tokens; one forward pass, nothing generated; run in process on 1x H100 80GB (Lium), quyet 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.07741166599589633,"p50_s_adjusted":0.3048233319917927,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"lev","name":"lev (Interfaze AI, Qwen3.5-4B + LoRA r32, label-token readout + candidate-path head, per-bucket temperatures)","source_url":"https://huggingface.co/interfaze-ai/lev","axes":{"intelligence":41.084702071234844,"calibration":82.32371666823741,"speed":88.00251499520694,"cost":61.79891447724963},"capability":{"score":61.70420936973613,"rank":18,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":42.177760514319104,"rank":14,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01876590517241379,"basis":"latest published v1.5 row (v1.5.8 combined file) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3); self-hosted estimate, not measured hosted billing"},"latency":{"p50_s_raw":0.10230199046782218,"p50_s_adjusted":0.3546039809356444,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"ninfer-qwen3.8-flash-next","name":"NInfer Qwen3.8-Flash-Next mixed","source_url":"https://github.com/igorls/ninfer","axes":{"intelligence":48.980301026876646,"calibration":90.37232453477458,"speed":89.33173525631216,"cost":42.53362878447125},"capability":{"score":69.67631278082561,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.5× Jev (v1.5 reference)"]},"composite_score":41.96785076957165,"rank":15,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.08232764778325123,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.08002880541607738,"p50_s_adjusted":0.3100576108321548,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jev-local","name":"jev-local (Qwen3.5-9B)","source_url":"https://github.com/us/jev-local","axes":{"intelligence":41.85041350559534,"calibration":73.06634984647151,"speed":75.55955847735143,"cost":58.85554343293177},"capability":{"score":57.458381676033426,"rank":23,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":41.328564334669615,"rank":16,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.023522413793103446,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.3653302099555731,"p50_s_adjusted":0.8806604199111462,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decider-4b-v2","name":"decider-4b v2 (Mapika)","source_url":"https://github.com/Mapika/decider","axes":{"intelligence":40.12597268917964,"calibration":89.63735637833331,"speed":91.78549329698745,"cost":64.54376873042901},"capability":{"score":64.88166453375648,"rank":9,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":41.240894350857424,"rank":17,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.015201077586206898,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.024963717442005873,"p50_s_adjusted":0.19992743488401174,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jevk5-v0.3-4b","name":"JevK5 v0.3 (4B)","source_url":"https://huggingface.co/alibiserikbay/JevK5","axes":{"intelligence":38.53201254374527,"calibration":89.88230801873149,"speed":93.9472917853559,"cost":63.07359395281681},"capability":{"score":64.20716028123837,"rank":10,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":37.36493548751909,"rank":18,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.018816601019352674,"p50_s_adjusted":0.18763320203870534,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"metask-jev-4b","name":"metask-jev-4b","source_url":"https://github.com/metask-ai/metask-jev","axes":{"intelligence":39.166454612935595,"calibration":84.41575352936586,"speed":90.78239008886905,"cost":57.7311327584882},"capability":{"score":61.79110407115073,"rank":17,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":37.34931073616377,"rank":19,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.025642610837438424,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.05397228497895412,"p50_s_adjusted":0.25794456995790827,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"deck-4b-v1-0","name":"deck-4B v1.0 (krishna765, JevK5 + LoRA, FP8 weight-only)","source_url":"https://github.com/krishna-gogineni-765/deck","axes":{"intelligence":38.62465958342338,"calibration":88.59116686443139,"speed":89.93848373337222,"cost":63.66277849512568},"capability":{"score":63.60791322392738,"rank":12,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":37.29399798821111,"rank":20,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0162645,"basis":"ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): DeepInfra Qwen/Qwen3.5-4B list price $0.03/M input, $0.15/M output (base of JevK5, the adapter's base checkpoint; not on OpenRouter; frozen 25 Sep snapshot per I-3, same reference as reflex-4b, jobe, typecastlm and Quyet) x the server's own measured usage.input_tokens; logit readout, 0 output tokens; FP8 weight-only on 1x RTX PRO 6000 96GB (Lium; v1.6 GPU-class deviation from source H100), deck-4b 1.0.0 + jevk5 f944fe37; estimated, not charged"},"latency":{"p50_s_raw":0.07620866177603602,"p50_s_adjusted":0.30241732355207207,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"clef-flash","name":"Clef-Flash (Cloudflare, Qwen3.5-9B post-train with a joint schema head, multimodal, measured on text)","source_url":"https://huggingface.co/Cloudflare/clef-flash","axes":{"intelligence":42.18368888957246,"calibration":85.46479352633332,"speed":88.4044196893156,"cost":46.79963593205177},"capability":{"score":63.824241207952895,"rank":11,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":36.635050479655426,"rank":21,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.05933971674876847,"basis":"latest published v1.5 row (v1.5.8 combined file) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied; self-hosted estimate, not measured hosted billing"},"latency":{"p50_s_raw":0.10283017950132489,"p50_s_adjusted":0.3556603590026498,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"qwen35-9b-jev-data-mix-v2","name":"Qwen3.5-9B Jev-like data-mix v2","source_url":"https://huggingface.co/jsaurabh/qwen3.5-9b-jev-data-mix-v2","axes":{"intelligence":41.94184606895763,"calibration":79.78895435560509,"speed":85.40793837616167,"cost":45.690218065856605},"capability":{"score":60.86540021228136,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.0× Jev (v1.5 reference)"]},"composite_score":33.58966697347588,"rank":22,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.06461391625615764,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.16828587953932583,"p50_s_adjusted":0.4865717590786517,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jevk5-v02","name":"JevK5 v0.2.0","source_url":"https://github.com/allebee/jevk5","axes":{"intelligence":36.261123811123895,"calibration":89.38710082819252,"speed":91.56570599843724,"cost":63.07359395281681},"capability":{"score":62.824112319658205,"rank":13,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":32.09901620353128,"rank":23,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.04039452798315324,"p50_s_adjusted":0.23078905596630647,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jevone","name":"JevOne","source_url":"https://huggingface.co/juspay/jev-one","axes":{"intelligence":45.955077428728934,"calibration":91.76581828544445,"speed":89.83885080776159,"cost":39.84212731591276},"capability":{"score":68.8604478570867,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 3.1× Jev (v1.5 reference)"]},"composite_score":31.14510331640707,"rank":24,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.10121908866995075,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.06179859535768628,"p50_s_adjusted":0.2735971907153726,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"hopper","name":"Hopper","source_url":"https://huggingface.co/HopitAI/hopper","axes":{"intelligence":34.73970870443837,"calibration":90.07902805935959,"speed":91.11434937226889,"cost":62.25627699896333},"capability":{"score":62.40936838189898,"rank":14,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":28.852652318563656,"rank":25,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.018118577586206896,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.0564777513500303,"p50_s_adjusted":0.26295550270006063,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"imajev-4b-rtx5090-a2","name":"Imajev-4B (RTX 5090)","source_url":"https://github.com/fstandhartinger/jevbench/issues/110","axes":{"intelligence":34.62714018309005,"calibration":89.1627567462182,"speed":90.51878000442085,"cost":63.26458220826236},"capability":{"score":61.89494846465412,"rank":15,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":28.655751132526984,"rank":26,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.016769261083743846,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (I-2): measured input tokens; zero generated output tokens for signed logits readout. M2 floor uses the 25 Sep 2026 DeepInfra Qwen3.5-4B snapshot rates (USD 0.03/M input, USD 0.15/M output); the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen3.5-9B."},"latency":{"p50_s_raw":0.07120202149963006,"p50_s_adjusted":0.29240404299926015,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"malkuth-4b","name":"Malkuth-4B (newfull5, Kev post-train)","source_url":"https://github.com/newfull5/malkuth","axes":{"intelligence":33.840703805987175,"calibration":88.43741591513763,"speed":89.51571180654442,"cost":55.82433433950372},"capability":{"score":61.139059860562405,"rank":19,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":26.197476126697616,"rank":27,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.02968398399014778,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.07427763848681934,"p50_s_adjusted":0.2985552769736387,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jqv","name":"jqv (Qwen3-32B zero-shot)","source_url":"https://github.com/Octalab-Inc/jqv","axes":{"intelligence":33.85815624201783,"calibration":89.87138320201578,"speed":83.1032423977654,"cost":51.172592607797476},"capability":{"score":61.8647697220168,"rank":16,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":25.391439220985372,"rank":28,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.042420935960591134,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.09740161104127765,"p50_s_adjusted":0.3448032220825553,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decider-35b-a3b","name":"decider-35b-a3b (Mapika)","source_url":"https://huggingface.co/Mapika/decider-35b-a3b","axes":{"intelligence":48.72471958402083,"calibration":83.61830413355554,"speed":91.27357492980725,"cost":34.38579876859529},"capability":{"score":66.1715118587882,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 4.8× Jev (v1.5 reference)"]},"composite_score":24.772887446577812,"rank":29,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.1538650708128079,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.04968994203954935,"p50_s_adjusted":0.2493798840790987,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"manchego-v21-v15","name":"Manchego v2.1","source_url":"https://github.com/nschlaepfer/manchego-serve","axes":{"intelligence":32.54303915979197,"calibration":84.0077997110931,"speed":90.7347510499745,"cost":64.32996474661243},"capability":{"score":58.27541943544253,"rank":22,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":24.487333472500133,"rank":30,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.015452586206896551,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: frozen 25 Sep DeepInfra Qwen/Qwen3.5-4B market reference, already marked deprecated at that cutoff (replaced by Qwen/Qwen3.5-9B). $0.03/M input, zero generated output; 836,500 measured input tokens across 1,624 decisions. Frozen v1.5 base-model floor applied; no bookable Manchego tariff claimed. Re-score if the basis changes."},"latency":{"p50_s_raw":0.05371422949974658,"p50_s_adjusted":0.2574284589994932,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decision-4b-v12","name":"Decision 4B v1.2 (FlyMyJev, Qwen3.5-4B + LoRA)","source_url":"https://benchmarkheaven.com/jev-models/decision-4b-v12","axes":{"intelligence":32.40310972936693,"calibration":88.42488519968094,"speed":93.92936299294371,"cost":63.07359395281681},"capability":{"score":60.41399746452394,"rank":20,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":24.463556699266523,"rank":31,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.019149313447996974,"p50_s_adjusted":0.18829862689599394,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"raw-qwen3-4b-instruct-2507","name":"Raw Qwen3 4B Instruct 2507 direct logits","source_url":"https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507","axes":{"intelligence":33.95117742100547,"calibration":35.18175191334879,"speed":91.16296741568695,"cost":63.36068112101963},"capability":{"score":34.56646466717713,"rank":53,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":21.792436038282993,"rank":32,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01664602832512315,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.04603758594021201,"p50_s_adjusted":0.24207517188042402,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"nimble-9b","name":"Bespoke Nimble 9B (Bespoke Labs)","source_url":"https://github.com/bespokelabsai/nimble","axes":{"intelligence":43.14237010986392,"calibration":69.88762704651819,"speed":86.11327010751721,"cost":36.75037423027614},"capability":{"score":56.514998578191054,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 4.0× Jev (v1.5 reference)"]},"composite_score":21.082616704957896,"rank":33,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.1283278325123153,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.14493575389496982,"p50_s_adjusted":0.43987150778993966,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"reflex-4b","name":"reflex 4B (kshetrajna12)","source_url":"https://github.com/kshetrajna12/reflex","axes":{"intelligence":30.935558296483705,"calibration":86.97925577932901,"speed":68.7900017304639,"cost":63.142712976263866},"capability":{"score":58.95740703790636,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 4.6× Jev (v1.5 reference)"]},"composite_score":20.637311182702398,"rank":34,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.016926853448275863,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":1.3577107690507546,"p50_s_adjusted":2.865421538101509,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"typecastlm","name":"typecastlm (Mikhail Gribov, Qwen3.5-4B computed head)","source_url":"https://github.com/mihail-gribov/typecastlm","axes":{"intelligence":29.78900656204624,"calibration":83.29951051808085,"speed":92.76354061232337,"cost":63.95900966831593},"capability":{"score":56.544258540063545,"rank":26,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":19.72256430004839,"rank":35,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01589887315270936,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.02951649052556604,"p50_s_adjusted":0.20903298105113208,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decision-4b-v11","name":"Decision 4B v1.1 (FlyMyJev, Qwen3.5-4B + LoRA)","source_url":"https://benchmarkheaven.com/jev-models/decision-4b-v11","axes":{"intelligence":29.134168280172865,"calibration":85.23934158755043,"speed":93.95641941218838,"cost":63.07359395281681},"capability":{"score":57.18675493386165,"rank":24,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":18.718356274892123,"rank":36,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.019026647554710507,"p50_s_adjusted":0.188053295109421,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"autojev-27b","name":"AutoJev-27B (denis-pplx, Qwen3.8-27B)","source_url":"https://huggingface.co/denis-pplx/autojev-27b","axes":{"intelligence":55.62422706919757,"calibration":87.61544276592178,"speed":88.60757446254297,"cost":29.365937020749882},"capability":{"score":71.61983491755967,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.0× Jev (v1.5 reference)"]},"composite_score":18.463427917707932,"rank":37,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.22618758620689655,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.09737715951632708,"p50_s_adjusted":0.3447543190326542,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"eikos-27b","name":"Eikos-27B (caiovicentino1, Qwen3.8-27B)","source_url":"https://huggingface.co/caiovicentino1/eikos-27b","axes":{"intelligence":64.40270761245412,"calibration":89.97714437568723,"speed":88.08630815059152,"cost":28.689752882469676},"capability":{"score":77.18992599407068,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.4× Jev (v1.5 reference)"]},"composite_score":18.077889022432696,"rank":38,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.23823646551724137,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.11261826660484076,"p50_s_adjusted":0.37523653320968153,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"ninfer-qwen3.8-27b","name":"NInfer Qwen3.8-27B NVFP4","source_url":"https://github.com/igorls/ninfer","axes":{"intelligence":51.800126951055375,"calibration":84.18557302696732,"speed":90.47734239273063,"cost":29.118887844204664},"capability":{"score":67.99284998901135,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.1× Jev (v1.5 reference)"]},"composite_score":17.71600496541158,"rank":39,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.2305174137931034,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.05501334450673312,"p50_s_adjusted":0.26002668901346626,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"openjev-thinking","name":"OpenJev (thinking, BF16)","source_url":"https://github.com/razorback16/openjev","axes":{"intelligence":70.34621384425424,"calibration":81.41345474033434,"speed":72.55270827201397,"cost":28.51480924396573},"capability":{"score":75.87983429229429,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.5× Jev (v1.5 reference)","latency 3.0× Jev (v1.5 reference)"]},"composite_score":17.265218966202397,"rank":40,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.24145693965517237,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.8569832294015214,"p50_s_adjusted":1.8639664588030427,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-jev-zefan-9b","name":"Open-Jev 9B (Zefan Cai)","source_url":"https://github.com/Zefan-Cai/Open-Jev","axes":{"intelligence":43.88591848753404,"calibration":79.80507069734225,"speed":74.39486349899005,"cost":33.05490478272432},"capability":{"score":61.84549459243814,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 5.3× Jev (v1.5 reference)","latency 2.2× Jev (v1.5 reference)"]},"composite_score":17.045611746249723,"rank":41,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.17041317733990147,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.6175056385109201,"p50_s_adjusted":1.3850112770218401,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"standardone-8b","name":"Standard One 8B (Standard Thinking)","source_url":"https://huggingface.co/StandardThinking/StandardOne-8B","axes":{"intelligence":32.80426604507879,"calibration":84.01508092906352,"speed":92.76442849252426,"cost":43.283250331904235},"capability":{"score":58.40967348707115,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.4× Jev (v1.5 reference)"]},"composite_score":16.917108299009165,"rank":42,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.07772456896551724,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.03173699742183089,"p50_s_adjusted":0.21347399484366178,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"clef","name":"Clef (Cloudflare, Qwen3.8-27B post-train with a joint schema head, multimodal, measured on text)","source_url":"https://huggingface.co/Cloudflare/clef","axes":{"intelligence":59.08165501447566,"calibration":91.57703482429629,"speed":83.3998209667487,"cost":28.102157220114762},"capability":{"score":75.32934491938597,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.7× Jev (v1.5 reference)"]},"composite_score":16.753692861922005,"rank":43,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.24922681034482758,"basis":"latest published v1.5 row (v1.5.8 combined file) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied; self-hosted estimate, not measured hosted billing"},"latency":{"p50_s_raw":0.21318734902888536,"p50_s_adjusted":0.5763746980577708,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"jev-qwen3.5-9b-base-nvfp4","name":"JEV Qwen3.5-9B Base NVFP4","source_url":"https://huggingface.co/WIlfLin/JEV-Qwen3.5-9B-Base-NVFP4","axes":{"intelligence":29.251131763478327,"calibration":85.03288206087805,"speed":93.62779491136789,"cost":47.592235403520306},"capability":{"score":57.14200691217819,"rank":25,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":15.97555620862577,"rank":44,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.05583743842364531,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (proxy tokens, I-2): exact base-model market reference"},"latency":{"p50_s_raw":0.02267279999796301,"p50_s_adjusted":0.19534559999592602,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"semif-qwen3.5-4b","name":"SemIf, formerly OpenJev (Qwen3.5-4B, TheoLeeCJ)","source_url":"https://github.com/TheoLeeCJ/openjev","axes":{"intelligence":27.10113669038628,"calibration":83.59960184071504,"speed":90.76560119421038,"cost":63.07359395281681},"capability":{"score":55.35036926555066,"rank":27,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":15.517179498801362,"rank":45,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.01701689039408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.05204228049842641,"p50_s_adjusted":0.25408456099685284,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"bev-bonsai-27b","name":"Bev / Bonsai 27B","source_url":"https://huggingface.co/Reza2kn/Bev","axes":{"intelligence":43.77513601921869,"calibration":88.50232899798883,"speed":72.98222461463935,"cost":28.231834531826323},"capability":{"score":66.13873250860377,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.6× Jev (v1.5 reference)","latency 3.4× Jev (v1.5 reference)"]},"composite_score":11.7393704396132,"rank":46,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.2467585344827586,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: documented hosted-model estimate. Frozen market reference qwen/qwen3.8-27b."},"latency":{"p50_s_raw":0.9861058448441327,"p50_s_adjusted":2.1222116896882652,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"litjev","name":"LitJev (Qwen3.8-27B)","source_url":"https://github.com/zhengxuyu/litjev","axes":{"intelligence":43.28549422942267,"calibration":89.9210925231847,"speed":68.17671225113584,"cost":28.35809912639418},"capability":{"score":66.60329337630368,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 7.6× Jev (v1.5 reference)","latency 5.0× Jev (v1.5 reference)"]},"composite_score":11.458886170296882,"rank":47,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.2443787068965517,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":1.4730074456892908,"p50_s_adjusted":3.0960148913785814,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"localjev-qwen3.5-4b","name":"local-jev Qwen3.5-4B","source_url":"https://github.com/amithgc/local-jev","axes":{"intelligence":24.145363218466134,"calibration":85.84286362718517,"speed":86.10068849774444,"cost":59.269627658320445},"capability":{"score":54.99411342282565,"rank":28,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":11.438161297684031,"rank":48,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.022786576354679802,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.1226705745793879,"p50_s_adjusted":0.39534114915877583,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"system-one-sg","name":"system-one (Qwen3-8B, Sean Goedecke)","source_url":"https://github.com/sgoedecke/system-one","axes":{"intelligence":29.271508779158296,"calibration":30.16473321941294,"speed":90.8867694864598,"cost":44.96855803039237},"capability":{"score":29.718120999285617,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.1× Jev (v1.5 reference)"]},"composite_score":11.027723483913586,"rank":49,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.06829381896551724,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.04996280049090274,"p50_s_adjusted":0.24992560098180547,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"kev-8b","name":"kev 8B (research preview)","source_url":"https://github.com/jaredpalmer/kev","axes":{"intelligence":29.26217941280341,"calibration":70.7561729087212,"speed":88.97143473610706,"cost":40.42054386073457},"capability":{"score":50.0091761607623,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 3.0× Jev (v1.5 reference)"]},"composite_score":10.62283914209564,"rank":50,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.09682374384236453,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.07448393944650888,"p50_s_adjusted":0.2989678788930178,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decision-2b","name":"Decision 2B (FlyMy.AI, v59)","source_url":"https://huggingface.co/flymy-ai/decision-2b-preview","axes":{"intelligence":22.875473256337628,"calibration":86.77033341356835,"speed":90.39462801231525,"cost":66.37275756963365},"capability":{"score":54.82290333495299,"rank":29,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":10.289708493744515,"rank":51,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.013210160098522166,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.07514283142518252,"p50_s_adjusted":0.30028566285036506,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"opensourcejev-qwen35-4b-q4km","name":"OpenSourceJev (Qwen3.5-4B Q4_K_M, native llama.cpp)","source_url":"https://github.com/sabeel111/OpenSourceJev","axes":{"intelligence":23.217995107845883,"calibration":73.92484912653177,"speed":77.06362388237805,"cost":69.13657649738097},"capability":{"score":48.571422117188824,"rank":33,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":10.263493713193125,"rank":52,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0106851539408867,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.29475566232576966,"p50_s_adjusted":0.7395113246515393,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-alternative-jev","name":"open-alternative-jev (Qwen3.5-4B, IkerMoel)","source_url":"https://github.com/ikermoel/open-alternative-jev","axes":{"intelligence":22.34610151162883,"calibration":75.94384178347806,"speed":91.30216983336405,"cost":63.27859776193043},"capability":{"score":49.14497164755345,"rank":32,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":9.435692571873393,"rank":53,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0167512315270936,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (proxy tokens, I-2): exact base-model market reference; reference deepinfra:Qwen/Qwen3.5-4B frozen at the 25 Sep cut-off; the snapshot records deprecation on 11 Jun 2026 and replacement by Qwen/Qwen3.5-9B; the frozen snapshot rates remain in use (I-3)"},"latency":{"p50_s_raw":0.04333395400317386,"p50_s_adjusted":0.2366679080063477,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"raw-qwen3-8b","name":"Raw Qwen3 8B direct logits","source_url":"https://huggingface.co/Qwen/Qwen3-8B","axes":{"intelligence":26.54912920590792,"calibration":33.721548452619835,"speed":88.65282990878164,"cost":45.53515592313027},"capability":{"score":30.13533882926388,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 2.0× Jev (v1.5 reference)"]},"composite_score":9.30125972789427,"rank":54,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0653875104679803,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.07753071398474276,"p50_s_adjusted":0.30506142796948554,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decider-2b","name":"decider-2b (Mapika)","source_url":"https://huggingface.co/Mapika/decider-2b","axes":{"intelligence":20.654576369710338,"calibration":68.52438190259257,"speed":94.55945724666857,"cost":64.91766177740826},"capability":{"score":44.58947913615145,"rank":37,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":7.6704332123718,"rank":55,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.014771046798029557,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.014745104941539466,"p50_s_adjusted":0.17949020988307893,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"nemotron-diffusion-8b","name":"Nemotron Diffusion 8B (pst2154, optimized vLLM)","source_url":"https://github.com/pst2154/Nemotron_Jev","axes":{"intelligence":20.38848344571935,"calibration":74.53740737677316,"speed":93.8305724848098,"cost":55.49170925809556},"capability":{"score":47.46294541124625,"rank":35,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":7.29748476516348,"rank":56,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.03045157019704434,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.02195067796856165,"p50_s_adjusted":0.1939013559371233,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"kev-4b","name":"kev 4B (research preview)","source_url":"https://github.com/jaredpalmer/kev","axes":{"intelligence":19.510707720813777,"calibration":68.45285936403559,"speed":90.46585183268016,"cost":65.78036497095258},"capability":{"score":43.98178354242468,"rank":39,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":6.611791819903415,"rank":57,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.013824661330049262,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.07199650770053267,"p50_s_adjusted":0.29399301540106537,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"malkuth-2b","name":"Malkuth-2B (newfull5, Kev post-train)","source_url":"https://github.com/newfull5/malkuth","axes":{"intelligence":15.917395195670291,"calibration":79.34885520197874,"speed":92.03462875001455,"cost":65.5661191189412},"capability":{"score":47.633125198824516,"rank":34,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":3.992169201455145,"rank":58,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.014053873152709357,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.04305429550004192,"p50_s_adjusted":0.23610859100008383,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"qwen3-reranker-4b","name":"Qwen3-Reranker-4B","source_url":"https://huggingface.co/Qwen/Qwen3-Reranker-4B","axes":{"intelligence":16.337944014530347,"calibration":73.15350658389404,"speed":82.14971164581814,"cost":48.39770052432463},"capability":{"score":44.7457252992122,"rank":36,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":3.715006798009075,"rank":59,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.05249000923645321,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: hosted exact-model reference (deepinfra:Qwen/Qwen3-Reranker-4B); no exact base-model floor applies"},"latency":{"p50_s_raw":0.20409131213091314,"p50_s_adjusted":0.5581826242618263,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"zerank-2","name":"ZeroEntropy zerank-2","source_url":"https://huggingface.co/zeroentropy/zerank-2-reranker","axes":{"intelligence":15.596611891228248,"calibration":83.20040692874471,"speed":82.89783918855927,"cost":48.39770052432463},"capability":{"score":49.39850940998648,"rank":31,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":3.349795360318651,"rank":60,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.05249000923645321,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: size-class proxy (deepinfra:Qwen/Qwen3-Reranker-4B); no exact base-model floor applies"},"latency":{"p50_s_raw":0.17457342706620693,"p50_s_adjusted":0.4991468541324139,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-jev-zefan-2b","name":"Open-Jev 2B (Zefan Cai)","source_url":"https://github.com/Zefan-Cai/Open-Jev","axes":{"intelligence":21.804441188912403,"calibration":73.61830158128652,"speed":77.13731150142542,"cost":33.05490478272432},"capability":{"score":47.71137138509946,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 5.3× Jev (v1.5 reference)"]},"composite_score":3.2383990318693976,"rank":61,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.17041317733990147,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":0.43650909198913723,"p50_s_adjusted":1.0230181839782744,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-jev-zefan-27b-v1.1","name":"Open-Jev 27B v1.1 (Zefan Cai, LoRA + decision head on Qwen3.8-27B)","source_url":"https://huggingface.co/ZefanCai/Open-Jev-27B-v1.1","axes":{"intelligence":52.45777412397041,"calibration":83.46836347105969,"speed":72.50871441156775,"cost":14.357426070787312},"capability":{"score":67.96306879751505,"rank":null,"eligible":false,"within_caps":false,"reasons":["cost 22.2× Jev (v1.5 reference)","latency 2.8× Jev (v1.5 reference)"]},"composite_score":2.8808671915746493,"rank":62,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.7157353448275862,"basis":"latest published v1.5 row (v1.5.8 combined file) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): Frozen BASE_REFERENCES Qwen/Qwen3.8-27B market price (USD 0.42/M input, 3.00/M output) applied to 2,767,510 input tokens and zero output tokens over 1,624 decisions; self-hosted estimate, not measured hosted billing. Lineage verified by architecture; no pricing gate. The author method encodes each candidate independently."},"latency":{"p50_s_raw":0.7902996460907161,"p50_s_adjusted":1.7305992921814322,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"raw-phi-4-mini","name":"Raw Phi-4 mini direct logits","source_url":"https://huggingface.co/microsoft/Phi-4-mini-instruct","axes":{"intelligence":13.660237978082595,"calibration":57.015465878675315,"speed":91.89255362835293,"cost":53.22051180195924},"capability":{"score":35.33785192837895,"rank":52,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":2.4794266829120515,"rank":63,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.03625068965517241,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.04178677382878959,"p50_s_adjusted":0.23357354765757918,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"simplejev-qwen3.5-0.8b","name":"SimpleJev (Qwen3.5-0.8B, CPU)","source_url":"https://github.com/featherless-ai/simple-jev","axes":{"intelligence":13.000426205687345,"calibration":50.197881904888476,"speed":58.929227255635794,"cost":70.31175519898045},"capability":{"score":31.59915405528791,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 14.3× Jev (v1.5 reference)"]},"composite_score":2.112084174105171,"rank":64,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.009763559113300492,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":4.323511580238119,"p50_s_adjusted":8.797023160476238,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"smalljev","name":"smalljev semantic-v9","source_url":"https://github.com/isHeSatoshi/smalljev","axes":{"intelligence":8.670533657732655,"calibration":77.23092648324078,"speed":90.21131723708834,"cost":60.77469204714053},"capability":{"score":42.95073007048671,"rank":43,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.7719455446134506,"rank":65,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.020300665024630543,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.06657194206491113,"p50_s_adjusted":0.2831438841298223,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"kev-0.6b","name":"kev 0.6B (research preview)","source_url":"https://github.com/jaredpalmer/kev","axes":{"intelligence":7.598789169415934,"calibration":66.26655558235618,"speed":91.51497362122684,"cost":80.09400261254245},"capability":{"score":36.93267237588606,"rank":51,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.5431213026023877,"rank":66,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004608220443349753,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.05674442206509411,"p50_s_adjusted":0.26348884413018825,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"opendecision","name":"OpenDecision (ModernBERT-large zero-shot)","source_url":"https://github.com/deepanwadhwa/OpenDecision","axes":{"intelligence":7.425225288997898,"calibration":68.56606448002526,"speed":87.12856375997052,"cost":79.10704862374335},"capability":{"score":37.99564488451158,"rank":48,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.508795165212109,"rank":67,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004970862068965518,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.07184981391765177,"p50_s_adjusted":0.29369962783530357,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"mghafiri-qwen3.5-0.8b-decision-model","name":"Qwen3.5-0.8B Decision Model (Mourad Ghafiri)","source_url":"https://huggingface.co/mghafiri/qwen3.5-0.8B-decision-model","axes":{"intelligence":7.297270616183145,"calibration":77.95657740696296,"speed":55.90747410573378,"cost":79.5180630553356},"capability":{"score":42.62692401157305,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 18.3× Jev (v1.5 reference)"]},"composite_score":0.47247471382875306,"rank":68,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.00481649630541872,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate"},"latency":{"p50_s_raw":5.551045611966401,"p50_s_adjusted":11.252091223932803,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"deem-0.8-v1","name":"Deem 0.8B v1","source_url":"https://huggingface.co/LibertAIDAI/deem-0.8-v1","axes":{"intelligence":6.531204174677637,"calibration":29.6278299611924,"speed":85.44505463939521,"cost":80.28784730320886},"capability":{"score":18.07951706793502,"rank":60,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.32342918922131697,"rank":69,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004540166256157636,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: documented hosted-model estimate. Frozen market reference deepinfra:Qwen/Qwen3.5-0.8B."},"latency":{"p50_s_raw":0.15738983592018485,"p50_s_adjusted":0.4647796718403697,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"decision-fast","name":"Decision Fast (FlyMy.AI, v53a)","source_url":"https://huggingface.co/flymy-ai/decision-fast-preview","axes":{"intelligence":6.110691078442272,"calibration":75.07737238688318,"speed":91.472185793536,"cost":80.08651867745036},"capability":{"score":40.59403173266273,"rank":45,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.2981489585308357,"rank":70,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004610868226600985,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.05768406600691378,"p50_s_adjusted":0.2653681320138276,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"quyet-1-0-small-en","name":"Quyet-1.0-Small-EN (Chinh Nguyen, ModernBERT-base encoder with fixed heads)","source_url":"https://huggingface.co/chinhnc/Quyet-1.0-Small-EN","axes":{"intelligence":6.054813340422714,"calibration":73.02134054125926,"speed":94.90893356488051,"cost":89.11296503596958},"capability":{"score":39.53807694084099,"rank":47,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.29239342204088814,"rank":71,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0023062222906403943,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): DeepInfra base-size encoders (bge-base, e5-base, gte-base, all-mpnet-base) $0.005/M input (ModernBERT-base has no market listing; base-size encoder reference) x the package's own measured usage.input_tokens; one forward pass, nothing generated; run in process on 1x H100 80GB (Lium), quyet 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.0145730719959829,"p50_s_adjusted":0.1791461439919658,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"fastino-gliner-2-5-decide","name":"Fastino GLiNER-2.5-Decide (hosted API)","source_url":"https://fastino.ai/","axes":{"intelligence":6.439988814479404,"calibration":68.83766854772081,"speed":84.7052606179839,"cost":46.59842351525712},"capability":{"score":37.63882868110011,"rank":49,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.28381932562808126,"rank":72,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.06026325,"basis":"ESTIMATE: published Fastino sources conflict (USD 0.03/M vs 0.15/M input); conservative higher USD 0.15/M input, 0/M output x measured tokens; no invoice"},"latency":{"p50_s_raw":0.4409300498664379,"p50_s_adjusted":0.4409300498664379,"adjustment":"none (API)"}},{"key":"laya-typed-decisions","name":"Laya typed-decisions","source_url":"https://huggingface.co/convaiinnovations/laya/tree/1c5edc17a7acd8701df6fc341c0d179f1c62c982/typed-decisions","axes":{"intelligence":5.471270906587037,"calibration":82.96996460403702,"speed":70.81965680870755,"cost":82.52942826196218},"capability":{"score":44.22061775531203,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 3.0× Jev (v1.5 reference)"]},"composite_score":0.21666109526429145,"rank":73,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.003822543103448276,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: documented hosted-model estimate; no exact base-model floor applies. $0.01/M input, $0/M output; same-class hosted-encoder estimate; no exact-base market floor."},"latency":{"p50_s_raw":0.8638670064901817,"p50_s_adjusted":1.8777340129803632,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"openjev-verdict-1.4","name":"openJev Verdict 1.4","source_url":"https://huggingface.co/heman10x/rlcd-modernbert-151m","axes":{"intelligence":5.035901218168153,"calibration":75.69890703057408,"speed":81.49274954261848,"cost":86.62313527343974},"capability":{"score":40.36740412437111,"rank":46,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.17222636557331997,"rank":74,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.002791871921182266,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: v1.5 measured-input proxy; no exact base-model floor applies"},"latency":{"p50_s_raw":0.2926154159940779,"p50_s_adjusted":0.7352308319881559,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"raw-qwen3-0.6b","name":"Raw Qwen3 0.6B direct logits","source_url":"https://huggingface.co/Qwen/Qwen3-0.6B","axes":{"intelligence":5.554864880982027,"calibration":10.94062892769364,"speed":93.52703208725202,"cost":77.58073177551512},"capability":{"score":8.247746904337834,"rank":63,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.1673536172919218,"rank":75,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.005588676108374384,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.02898204093798995,"p50_s_adjusted":0.2079640818759799,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"lev-350m","name":"lev-350m (Franck Verrot, LFM2.5-350M)","source_url":"https://github.com/franckverrot/lev","axes":{"intelligence":4.827562367090417,"calibration":81.86424479313416,"speed":93.98230372494703,"cost":80.03797760430007},"capability":{"score":43.34590358011229,"rank":41,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.15377137469809216,"rank":76,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.00462807881773399,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.022326925001834752,"p50_s_adjusted":0.1946538500036695,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"raw-qwen3-1.7b","name":"Raw Qwen3 1.7B direct logits","source_url":"https://huggingface.co/Qwen/Qwen3-1.7B","axes":{"intelligence":5.124286554530066,"calibration":11.516326812547591,"speed":93.23530725162705,"cost":68.54983190559568},"capability":{"score":8.320306683538828,"rank":62,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.13671939226506033,"rank":77,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.011177352216748768,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.03198186121881008,"p50_s_adjusted":0.21396372243762016,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"kev-0.5b","name":"kev 0.5B","source_url":"https://github.com/jaredpalmer/kev","axes":{"intelligence":4.598433868646836,"calibration":61.20618525995795,"speed":92.39904218127765,"cost":80.09398520305423},"capability":{"score":32.90230956430239,"rank":54,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.131588623265787,"rank":78,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004608226600985222,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.043983193347230554,"p50_s_adjusted":0.2379663866944611,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"openjev-verdict","name":"openJev Verdict (heman10x, ModernBERT-base 151M)","source_url":"https://github.com/Heman10x-NGU/openJev-verdict-2.0","axes":{"intelligence":4.1496399816416485,"calibration":49.11476555745841,"speed":79.40046783928,"cost":86.62313527343974},"capability":{"score":26.63220276955003,"rank":58,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.09650709186186587,"rank":79,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.002791871921182266,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: v1.5 measured-input proxy; no exact base-model floor applies"},"latency":{"p50_s_raw":0.2692721150815487,"p50_s_adjusted":0.6885442301630974,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"quyet-1-0-small","name":"Quyet-1.0-Small (Chinh Nguyen, SEA-LION-ModernBERT-300M encoder with fixed heads)","source_url":"https://huggingface.co/chinhnc/Quyet-1.0-Small","axes":{"intelligence":4.019045994532796,"calibration":60.96568104722222,"speed":94.91846454857637,"cost":79.56261245432609},"capability":{"score":32.492363520877504,"rank":55,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.08963716950381444,"rank":80,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.004800055418719211,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): DeepInfra same-size encoders (bge-large, e5-large, Qwen3-Embedding-0.6B) $0.01/M input (SEA-LION-ModernBERT-300M has no market listing; the ModernBERT-large-class reference of the Laya/Certo/OpenDecision rows) x the package's own measured usage.input_tokens; one forward pass, nothing generated; run in process on 1x H100 80GB (Lium), quyet 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.01451935400837101,"p50_s_adjusted":0.17903870801674202,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"quyet-1-0-tiny","name":"Quyet-1.0-Tiny (Chinh Nguyen, mmBERT-small encoder (16 layers kept) with fixed heads)","source_url":"https://huggingface.co/chinhnc/Quyet-1.0-Tiny","axes":{"intelligence":3.6964771541368573,"calibration":58.46458026762962,"speed":95.09610910731021,"cost":88.43590631389968},"capability":{"score":31.08052871088324,"rank":56,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.07064766318734213,"rank":81,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0024292364532019703,"basis":"v1.5.8 fast-lane delivery row (combined v1.5.8 preview, not yet published) cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE (v1.5-M2 base-model reference; self-hosted open weights, no public tariff): DeepInfra base-size encoders (bge-base, e5-base, gte-base, all-mpnet-base) $0.005/M input (mmBERT-small 16-layer, 183M; base-size class reference, errs high vs the small-encoder $0.004 class) x the package's own measured usage.input_tokens; one forward pass, nothing generated; run in process on 1x H100 80GB (Lium), quyet 1.0.0; estimated, not charged"},"latency":{"p50_s_raw":0.012666817499848548,"p50_s_adjusted":0.1753336349996971,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-jev-deberta-v3-large","name":"open-jev-deberta-v3-large (local CPU)","source_url":"https://github.com/kotoba-lang/typed-decisions","axes":{"intelligence":2.8293414417857914,"calibration":67.96597988813339,"speed":67.62138193768119,"cost":77.5922354035203},"capability":{"score":35.39766066495959,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 5.7× Jev (v1.5 reference)"]},"composite_score":0.03235814777127837,"rank":82,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.005583743842364532,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: v1.5 measured-input proxy; no exact base-model floor applies"},"latency":{"p50_s_raw":1.6890084934420884,"p50_s_adjusted":3.5280169868841766,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"verdict-small","name":"verdict-small (Manavarya09, multilingual-e5-small 118M)","source_url":"https://github.com/Manavarya09/verdict","axes":{"intelligence":2.3417095945172277,"calibration":72.92000413192591,"speed":89.48731594094681,"cost":100},"capability":{"score":37.63085686322157,"rank":50,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.01899384266854915,"rank":83,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.000866152709359606,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.04910206887871027,"p50_s_adjusted":0.24820413775742053,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"laya","name":"Laya (Convai Innovations, ModernBERT-large 421M)","source_url":"https://huggingface.co/convaiinnovations/laya","axes":{"intelligence":1.7910341507793501,"calibration":63.22326564159259,"speed":73.26302890816496,"cost":84.89402784895724},"capability":{"score":32.50714989618597,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 2.9× Jev (v1.5 reference)"]},"composite_score":0.008560099345041717,"rank":84,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0031881034482758625,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.825308452360332,"p50_s_adjusted":1.800616904720664,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"mxbai-rerank-base-v2","name":"Mixedbread mxbai-rerank-base-v2","source_url":"https://huggingface.co/mixedbread-ai/mxbai-rerank-base-v2","axes":{"intelligence":1.140707327109041,"calibration":86.60772459153094,"speed":92.58758170313658,"cost":60.33589314238483},"capability":{"score":43.87421595931999,"rank":40,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.002273929358844764,"rank":85,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.020996016009852216,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: size-class proxy (deepinfra:Qwen/Qwen3-Reranker-0.6B); no exact base-model floor applies"},"latency":{"p50_s_raw":0.031425865134224296,"p50_s_adjusted":0.21285173026844859,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"laya-multilingual","name":"Laya multilingual","source_url":"https://huggingface.co/convaiinnovations/laya/tree/1c5edc17a7acd8701df6fc341c0d179f1c62c982/multilingual","axes":{"intelligence":0.2879184668737643,"calibration":33.655846666999985,"speed":77.82044954853706,"cost":82.1551651086044},"capability":{"score":16.971882566936873,"rank":61,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.000037595668430455566,"rank":86,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.003933940886699508,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: documented hosted-model estimate; no exact base-model floor applies. $0.01/M input, $0/M output; same-class hosted-encoder estimate; no exact-base market floor."},"latency":{"p50_s_raw":0.3577279045130126,"p50_s_adjusted":0.8654558090260253,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"certo","name":"Certo v1 (AltSlate Labs)","source_url":"https://huggingface.co/altslate/certo-decision-model","axes":{"intelligence":0.20278658812856143,"calibration":86.14595667337036,"speed":92.53493469610669,"cost":96.9130801359906},"capability":{"score":43.17437163074946,"rank":42,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.000013254531668229452,"rank":87,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0012673522167487683,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.042036901926621795,"p50_s_adjusted":0.23407380385324358,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"clm-8b","name":"CLM-8B (Contrastive-LM, clm-latest)","source_url":"https://github.com/Contrastive-LM/CLM","axes":{"intelligence":0.13075619159747265,"calibration":40.041900621338065,"speed":92.63548383949004,"cost":50.501662240594015},"capability":{"score":20.086328406467768,"rank":59,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0.0000035510969933114796,"rank":88,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.044662660714285714,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): price floor: base-model reference price applied"},"latency":{"p50_s_raw":0.02119073923677206,"p50_s_adjusted":0.19238147847354412,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"bge-reranker-v2-m3","name":"BAAI bge-reranker-v2-m3","source_url":"https://huggingface.co/BAAI/bge-reranker-v2-m3","axes":{"intelligence":0.06512222845080358,"calibration":88.10046932372114,"speed":93.86976236118387,"cost":59.325605115070715},"capability":{"score":44.08279577608597,"rank":38,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":4.4076803151681546e-7,"rank":89,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.022688885467980296,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: base-model market reference (deepinfra:BAAI/bge-m3)"},"latency":{"p50_s_raw":0.01707328436896205,"p50_s_adjusted":0.1841465687379241,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"gte-reranker-modernbert-base","name":"Alibaba GTE Reranker ModernBERT-base","source_url":"https://huggingface.co/Alibaba-NLP/gte-reranker-modernbert-base","axes":{"intelligence":0,"calibration":83.4556233651092,"speed":93.76120791185616,"cost":69.26965431492843},"capability":{"score":41.7278116825546,"rank":44,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0,"rank":90,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.010576570197044334,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): ESTIMATE: size-class proxy (deepinfra:thenlper/gte-base); no exact base-model floor applies"},"latency":{"p50_s_raw":0.024179050931707025,"p50_s_adjusted":0.19835810186341404,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"mirror","name":"Mirror","source_url":"https://github.com/Bluusun/Decision-API","axes":{"intelligence":0,"calibration":51.61277239297377,"speed":73.47211485993975,"cost":89.27300805786774},"capability":{"score":25.806386196486883,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 2.4× Jev (v1.5 reference)"]},"composite_score":0,"rank":91,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.0022780665024630543,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.6508294129744172,"p50_s_adjusted":1.4516588259488343,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"open-jev-json-canvas-joshuasp","name":"Open Jev JSON Canvas (JoshuaSP)","source_url":"https://github.com/JoshuaSP/open-jev","axes":{"intelligence":61.19221807027989,"calibration":0,"speed":85.53332924416175,"cost":49.295219536461964},"capability":{"score":30.596109035139946,"rank":57,"eligible":true,"within_caps":true,"reasons":[]},"composite_score":0,"rank":92,"ranked":true,"listing":"ranked","not_ranked_because":null,"price":{"kind":"estimate","usd_per_1000_decisions":0.04899585591133005,"basis":"v1.5.4 published cost per 1,000 decisions carried (pricing rules unchanged; v1.6 item lengths differ): documented hosted-model estimate; no exact base-model floor applies"},"latency":{"p50_s_raw":0.14833511249162257,"p50_s_adjusted":0.44667022498324516,"adjustment":"x2 + 0.15 s (assumption, not measured)"}},{"key":"wity-1-always","name":"wity-1 (Wity, reasoning always)","source_url":"https://wity.alphanimble.com/","axes":{"intelligence":70.45905034996007,"calibration":90.96578585343975,"speed":71.20296182763504,"cost":58.368182152360596},"capability":{"score":80.71241810169991,"rank":null,"eligible":false,"within_caps":false,"reasons":["latency 2.7× Jev (v1.5 reference)"]},"composite_score":70.96883700634847,"rank":null,"ranked":false,"listing":"listed","not_ranked_because":"registry ranked=false or no composite","price":{"kind":"estimate","usd_per_1000_decisions":0.02441896551724138,"basis":"Wity's stated API tariff: USD 0.042 per 1M input tokens, output free; v1.5 cost per 1,000 decisions carried from the v1.5.8 build (measured 944,200 input tokens over 1,624 decisions, build f65ce826a455). API models with a known base are ranked at their own API price (1 Oct 2026 rule)."},"latency":{"p50_s_raw":1.6885605081915855,"p50_s_adjusted":1.6885605081915855,"adjustment":"none (API)"}},{"key":"wity-1-off","name":"wity-1 (Wity, reasoning off)","source_url":"https://wity.alphanimble.com/","axes":{"intelligence":40.91725354389175,"calibration":74.90911730698414,"speed":89.55912090143326,"cost":58.368182152360596},"capability":{"score":57.913185425437945,"rank":null,"eligible":false,"within_caps":true,"reasons":[]},"composite_score":40.53344853152938,"rank":null,"ranked":false,"listing":"listed","not_ranked_because":"registry ranked=false or no composite","price":{"kind":"estimate","usd_per_1000_decisions":0.02441896551724138,"basis":"Wity's stated API tariff: USD 0.042 per 1M input tokens, output free; v1.5 cost per 1,000 decisions carried from the v1.5.8 build (measured 944,200 input tokens over 1,624 decisions, build f65ce826a455). API models with a known base are ranked at their own API price (1 Oct 2026 rule)."},"latency":{"p50_s_raw":0.28173011541366577,"p50_s_adjusted":0.28173011541366577,"adjustment":"none (API)"}}]}
|