whatmodel.sh / decision / jev-brain · Gemma-4-31B (decision readout)

jev-brain · Gemma-4-31B (decision readout)

+ contribute

jev-brain · Gemma-4-31B (decision readout) is a source-available decision implementation from wcole3. Its published size is 31.27B parameters.

Catalogue identity notes

Exact Decision Index tested engine jev-brain-gemma-4-31b; reported model jev-brain · Gemma-4-31B. Type: inference technique; technique: autoregressive. Implementation: https://github.com/wcole3/jev-brain/tree/di-0.3-gemma-4-31b-r2. Comparison entry; no current run command, API price or speed inferred from the benchmark.

Model information sources

· Parameter count ↗

Publisher parameter count; includes the served backbone, not just adapter size.

64.5±1.4#59 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of jev-brain · Gemma-4-31B (decision readout), out of 357 with a score.

#	model	score	where	RAM	released
57	Jebadiah 27B v2.1 (T1e) frontier-infra	65±1	local	~69GB	2026-10
58	OneJev1.5-27B OmniJev	65±4	model files		2026-10-13
59	jev-brain · Gemma-4-31B (decision readout) wcole3	65±1	source code
60	Aleria Judge 1.1.0 (Gemma 4 31B) (decision readout) Decision Index submission: aleria-judge-1.1.0	65±4	source code		2026-10-14
61	d1 (hosted) Liquid AI	65±1	api price unknown

see all decision models · find the ones that fit your machine

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #51 of 250 · 6.1% of its score · 6 tests

Beats vLLM-SR d3-ultra (Decision 3.0, 27B), AutoTrust-JEV-27B v1.1, OneShot 27B here, though they rank higher overall. Trails baselabs joey-1 (2B), Solomon v1.1, Bespoke Nimble 9B v3 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: API-Bank (accuracy) submitted	85.0% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 160/236	easy · avg 70%	▇ 1.8%	sharp · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	62.5% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 149/243	medium · avg 37%	▇ 1.1%	sharp · at its level · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	66.9% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 182/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	97.4% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 198/246	very easy · avg 92%	▇ 1.1%	sharp · maxed out · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	79.1% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 133/246	easy · avg 72%	▇ 1.0%	noisy
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	79.9% ✎	no other model yet		—	failed run: Publisher excludes this protocol because its prompt reveals GPT-4 as the best route. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json

retrieval & classification▁▁▁·▁▁▁▁▃▃█▅ #70 of 258 · 7.5% of its score · 7 tests

Beats AutoTrust-JEV-27B v1.1, oenogarum (Gemma 4 31B), Decider 31B (NVFP4) here, though they rank higher overall. Trails StartLux-Decision v1.1 0.8B, Standard One 8B SH (v4), StartLux-Decision v1.1 2B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	48.1% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 199/246	medium · avg 38%	▇ 3.0%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	57.1% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 181/246	medium · avg 47%	▇ 2.7%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	76.6% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 130/246	easy · avg 74%	▇ 0.6%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	96.9% ✎	█▅▅▄▃▂▁▁▃▁·▁ beats 238/245	easy · avg 70%	▇ 0.6%	noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	78.9% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 83/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	90.8% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 139/238	easy · avg 83%	▇ 0.3%	sharp · maxed out · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	6.7% ✎	no other model yet		—	failed run: Publisher reports a builder bug exposing dialogue turns after the classified turn; awaiting corrected rerun. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json

language understanding▂▂▂▃▂▅▆▃█▆▄▃ #113 of 256 · 22% of its score · 9 tests

Trails Decider 4B, jiwo 4B v2, Sieve 4B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: ANLI (macro-F1) submitted	78.5% ✎	▁▂▃▅▅█▃····▁ beats 225/246	medium · avg 60%	▇▇ 5.5%	at its level
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	80.8% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 159/246	easy · avg 74%	▇▇ 5.2%	typical
Decision Index 0.4.1: HellaSwag (accuracy) submitted	91.9% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 139/246	very easy · avg 85%	▇ 3.3%	sharp · maxed out
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	59.6% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 182/246	medium · avg 44%	▇ 2.7%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	40.2% ✎	▁··▂▃▃▅▇█▇▃▃ beats 200/246	hard · avg 22%	▇ 1.7%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	78.6% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 205/246	medium · avg 62%	▇ 1.1%	at its level
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	78.4% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 148/246	easy · avg 71%	▇ 1.0%	noisy
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	77.9% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 157/246	easy · avg 66%	▇ 0.7%	noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: FinEntity (macro-F1) submitted	93.4% ✎	▁···▁▁▁▁▃▆█▃ beats 203/246	easy · avg 84%	▇ 0.5%	maxed out · noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #97 of 247 · 10% of its score · 6 tests

Beats Jev 1.13 here, though they rank higher overall. Trails vLLM-SR vela3-fast-vision (Vela 3.0, 0.25B), Crivo-400M, Amber 0.4B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: New Yorker (accuracy) submitted	77.5% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 205/246	medium · avg 61%	▇▇ 4.1%	typical
Decision Index 0.4.1: Humicroedit (accuracy) submitted	62.9% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 168/246	medium · avg 60%	▇ 2.3%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	37.9% ✎	▄▅▅▃▆▅▃▅▃▅▃█ beats 157/237	hard · avg 21%	▇ 1.3%	sharp · at its level · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	65.8% ✎	▁········▁▇█ beats 202/246	medium · avg 60%	▇ 1.3%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	90.6% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 147/246	easy · avg 83%	▇ 0.9%	sharp · maxed out · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	48.4% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 170/246	medium · avg 47%	▇ 0.4%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #46 of 263 · 43% of its score · 14 tests

Beats OneJev1.6-27B, baselabs elephant-1 (9B), StartLux-Decision v1.1 9B here, though they rank higher overall. Trails Solar Decide, Clef-omni, GPT-6 Luna Decisions here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MMLU-Pro (accuracy) submitted	69.1% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 201/246	medium · avg 46%	▇▇▇▇▇ 11%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	87.0% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 209/246	easy · avg 70%	▇▇ 6.1%	sharp
Decision Index 0.4.1: BBH (accuracy) submitted	75.1% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 156/246	easy · avg 66%	▇▇ 6.0%	typical
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	48.5% ✎	▂▆▅▅▆█▆▃▁··▁ beats 182/246	medium · avg 39%	▇▇ 3.9%	blunt · at its level
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	73.2% ✎	▁··▄█▆▆▆▅▅▃▁ beats 198/246	medium · avg 45%	▇ 3.7%	sharp · at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	87.5% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 178/246	easy · avg 78%	▇ 2.8%	typical
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	97.9% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 240/246	very easy · avg 90%	▇ 2.1%	sharp · maxed out
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	34.5% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 198/246	hard · avg 22%	▇ 2.0%	at its level

6 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MuSR (accuracy) submitted	68.5% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 203/246	medium · avg 60%	▇ 1.4%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	80.3% ✎	██▄▃▃▄▁▁···▁ beats 197/246	medium · avg 58%	▇ 1.2%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	81.9% ✎	▁███▆▅▃▃▁▁▁▁ beats 186/246	easy · avg 71%	▇ 1.0%	typical
Decision Index 0.4.1: ChessBench (accuracy) submitted	23.2% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 204/246	hard · avg 16%	▇ 0.9%	blunt
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	99.0% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 197/246	very easy · avg 96%	▇ 0.9%	sharp · maxed out · noisy
Decision Index 0.4.1: HLE (accuracy) submitted	14.8% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 190/246	very hard · avg 13%	▇ 0.0%	blunt

other tests2 not in a task · 10% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	61.2200 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 213/247	avg 41.206 points	▇▇▇ 6.6%	blunt
Decision Index 0.4.1: private new domains (equated points)	69.7000 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 212/247	avg 38.861 points	▇ 3.6%	blunt

$ where to run it

Implementation and setup instructions ↗
Published implementation and setup documentation verified. This entry describes a decision readout or evaluated configuration, not necessarily separate trained weights. Follow the referenced version and model settings; hardware compatibility has not been verified. Source · checked 2026-10-31.

Model repository and documentation: github.com/wcole3 ↗

$ how the score was worked out · Decision Index 0.4.1 99%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Decision Index 0.4.1 · submitted; independent execution unverified	42	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 64.5

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	79.9%	failed run: Publisher excludes this protocol because its prompt reveals GPT-4 as the best route. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json
Decision Index 0.4.1	Decision Index 0.4.1: SGD/SGD-X (macro-F1) [correct]	6.7%	failed run: Publisher reports a builder bug exposing dialogue turns after the classified turn; awaiting corrected rerun. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-11-03 02:37 UTC · sources and licences · json