JPT-35B-A3B

+ contribute

JPT-35B-A3B is an open-weights decision model from kirp. Its published size is 35.95B parameters. Licence: cc-by-nc-4.0.

Catalogue identity notes

Exact Decision Index tested engine jpt-35b-a3b; reported model JPT-35B-A3B. Type: LoRA; technique: autoregressive. Comparison entry; no current run command, API price or speed inferred from the benchmark. Current repository metadata checked separately; an unpinned alias is not evidence that today’s weights reproduce this historical run.

Model information sources

· Weights licence ↗ · Parameter count ↗

Publisher parameter count (estimated by publisher); includes the served backbone, not just adapter size.

58.2±1.3#126 of 355 decision modelsfrom 44 benchmark results across 2 sources

$ how it compares

The decision models ranked either side of JPT-35B-A3B, out of 357 with a score, then more from kirp.

#	model	score	where	RAM	released
126	vLLM-SR d3-mini (4B) vLLM Semantic Router	58±1	local	~11GB
127	Seb-9B v2 ironbcc	58±1	model files
128	JPT-35B-A3B kirp	58±1	model files
129	Mica v0.5 4B akivet	58±1	model files		2026-10
130	AJev lora7 (Gemma 4 12B) andyzhang232	58±1	model files
	more from kirp
178	JPT-9B Qwen3.5-9B LoRA kirp	54±1	model files
204	JPT-4B Qwen3.5-4B LoRA kirp	50±1	model files
289	JPT-0.8B Qwen3.5-0.8B LoRA kirp	33±1	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #100 of 250 · 6.8% of its score · 6 tests

Beats Jev 1.13, Xor 26B-A4B v1.2, Senova-4B v2 here, though they rank higher overall. Trails OneJev1.6-0.8B, TokenRhythm NeoHorse-Jev-4B, mmastrac diffgemma (decision readout) 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)	85.6% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 181/236	easy · avg 70%	▇ 2.1%	sharp · noisy
Decision Index 0.4.1: BFCL (case exact accuracy)	96.0% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 146/246	very easy · avg 92%	▇ 1.5%	sharp · maxed out · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy)	34.1% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 109/243	medium · avg 37%	▇ 1.2%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy)	74.5% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 111/246	easy · avg 72%	▇ 1.0%	noisy
Decision Index 0.4.1: ToolRet (nDCG@10)	63.4% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 108/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective))	80.0% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #124 of 258 · 6.6% of its score · 7 tests

Beats AutoTrust-JEV-27B v1.1, Quyet-1.0-Large, Tern-35B-A3B here, though they rank higher overall. Trails vLLM-SR vela3-fast (Vela 3.0, 0.4B), crystal-0.8b, Kas 4B 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)	44.7% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 149/246	medium · avg 38%	▇ 2.5%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1)	53.1% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 130/246	medium · avg 47%	▇ 2.3%	at its level
Decision Index 0.4.1: HoVer (accuracy)	67.7% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 90/246	easy · avg 74%	▇ 0.6%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy)	75.5% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 156/245	easy · avg 70%	▇ 0.5%	noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1)	85.0% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 78/238	easy · avg 83%	▇ 0.4%	sharp · noisy
Decision Index 0.4.1: BANKING77 (macro-F1)	77.3% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 71/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1)	59.8% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #115 of 256 · 22% of its score · 9 tests

Beats garum (Gemma 4 31B), Shisa DE-2.1 (repeat), Blink v0.3 26B-A4B (NVFP4) here, though they rank higher overall. 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)	71.3% ✎	▁▂▃▅▅█▂····▁ beats 168/246	medium · avg 60%	▇▇ 5.2%	at its level
Decision Index 0.4.1: NLI4CT (macro-F1)	82.4% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 180/246	easy · avg 74%	▇▇ 5.2%	typical
Decision Index 0.4.1: HellaSwag (accuracy)	91.8% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 138/246	very easy · avg 85%	▇▇ 4.4%	sharp · maxed out
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English)	61.0% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 190/246	medium · avg 44%	▇ 2.4%	at its level
Decision Index 0.4.1: ACOS (per-review F1)	21.1% ✎	▁··▂▃▃▅▇█▇▃▃ beats 117/246	hard · avg 22%	▇ 1.3%	at its level · noisy
Decision Index 0.4.1: ContractNLI (macro-F1)	81.7% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 179/246	easy · avg 71%	▇ 1.0%	noisy
Decision Index 0.4.1: VAST (macro-F1)	68.3% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 127/246	medium · avg 62%	▇ 1.0%	at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class)	75.7% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 133/246	easy · avg 66%	▇ 0.8%	at its level · noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: FinEntity (macro-F1)	93.4% ✎	▁···▁▁▁▁▃▆█▃ beats 202/246	easy · avg 84%	▇ 0.6%	noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #114 of 247 · 9.4% of its score · 6 tests

Beats Bobcat 1.3 (31B), Nace.AI Drex v2.0 (35B-A3B), decisio 0.6.0 (Gemma 4 31B) (decision readout) 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)	69.1% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 147/246	medium · avg 61%	▇▇ 3.9%	at its level
Decision Index 0.4.1: Humicroedit (accuracy)	61.7% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 141/246	medium · avg 60%	▇ 1.9%	blunt · at its level
Decision Index 0.4.1: cfcolor (accuracy)	62.3% ✎	▁········▁▇█ beats 132/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy)	89.5% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 133/246	easy · avg 83%	▇ 1.1%	sharp · noisy
Decision Index 0.4.1: POP909-CL (accuracy)	12.3% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 108/237	hard · avg 21%	▇ 1.0%	sharp · at its level · noisy
Decision Index 0.4.1: Habermas Machine (accuracy)	43.7% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 115/246	medium · avg 47%	▇ 0.3%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #93 of 263 · 43% of its score · 14 tests

Beats StartLux-Decision v1.1 4B, KnowLine 4B Gen5.5, Jeffan-4B-2010 here, though they rank higher overall. Trails Solar Decide, Clef-flash 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)	61.7% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 166/246	medium · avg 46%	▇▇▇▇ 11%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy)	84.6% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 183/246	easy · avg 70%	▇▇▇ 6.5%	sharp
Decision Index 0.4.1: BBH (accuracy)	72.7% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 150/246	easy · avg 66%	▇▇ 6.1%	typical
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice))	40.8% ✎	▁··▄█▆▆▆▅▅▃▁ beats 113/246	medium · avg 45%	▇ 3.5%	sharp · at its level
Decision Index 0.4.1: GPQA Diamond (accuracy)	48.0% ✎	▂▆▅▅▆█▆▃▁··▁ beats 173/246	medium · avg 39%	▇ 3.3%	blunt · at its level
Decision Index 0.4.1: ARC-Challenge (accuracy)	96.3% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 168/246	very easy · avg 90%	▇ 3.1%	sharp · maxed out
Decision Index 0.4.1: WinoGrande (accuracy)	83.0% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 140/246	easy · avg 78%	▇ 2.9%	typical
Decision Index 0.4.1: SATA-Bench (case exact accuracy)	22.4% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 86/246	hard · avg 22%	▇ 1.5%	at its level

6 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: ARC-Easy (accuracy)	99.0% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 176/246	very easy · avg 96%	▇ 1.5%	sharp · maxed out · noisy
Decision Index 0.4.1: MuSR (accuracy)	58.8% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 119/246	medium · avg 60%	▇ 1.2%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy)	67.0% ✎	██▄▃▃▄▁▁···▁ beats 159/246	medium · avg 58%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy)	69.4% ✎	▁███▆▅▃▃▁▁▁▁ beats 125/246	easy · avg 71%	▇ 0.9%	typical
Decision Index 0.4.1: ChessBench (accuracy)	16.8% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 130/246	hard · avg 16%	▇ 0.7%	blunt
Decision Index 0.4.1: HLE (accuracy)	11.4% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 74/246	very hard · avg 13%	▇ 0.0%	blunt

other tests4 not in a task · 12% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	51.8000 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 153/247	avg 41.206 points	▇▇ 5.5%	blunt
Decision Index 0.4.1: private new domains (equated points)	57.5200 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 177/247	avg 38.861 points	▇ 3.0%	blunt
Benchmark Heaven jevbench v1.6.1: intelligence (points)	51.4221 points ✎	▃···▃·▃▃█▅▅█ beats 3/13	avg 32.166 points	▇ 2.4%	blunt
Benchmark Heaven jevbench v1.6.1: calibration (points)	89.9314 points ✎	▂······▂▆▂██ beats 8/13	avg 70.133 points	▇ 0.7%	blunt

$ where to run it

Download model files ↗
Published model files verified. This is a verified repository snapshot; the benchmark does not establish an identical weight revision. A local run command and hardware requirements have not been verified for this entry. Source · checked 2026-10-31.

Model repository and documentation: huggingface.co/kirp ↗

$ how the score was worked out · Decision Index 0.4.1 96%, Benchmark Heaven (own decisio… 3%

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	42	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 96%	CC BY 4.0
Benchmark Heaven (own decision evaluations)	2	▇ 3%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 58.2

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	80.0%	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]	59.8%	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