whatmodel.sh / decision / StartLux-Decision v1.1 2B

StartLux-Decision v1.1 2B

+ contribute

StartLux-Decision v1.1 2B is an open-weights decision model from StartLux. Its published size is 2.27B parameters. Licence: cc-by-nc-4.0.

Catalogue identity notes

Exact Decision Index tested engine startlux-decision-v1.1-2b-instant; reported model StartLux-Decision v1.1 2B. Type: full fine-tune; 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; includes the served backbone, not just adapter size.

55.3±1.2#156 of 355 decision modelsfrom 50 benchmark results across 2 sources

$ how it compares

The decision models ranked either side of StartLux-Decision v1.1 2B, out of 357 with a score, then more from StartLux.

#	model	score	where	RAM	released
156	Decider 35B-A3B Mapika	55±1	model files		2026-09-20
157	Amber 4B MLukauskas	55±1	model files
158	StartLux-Decision v1.1 2B StartLux	55±1	model files
159	kev-4b (DecisionBench 485ace87) Jared Palmer	55±2	model files
160	Jebadiah 9B v2.1 (Ollaya Q8_0) frontier-infra	55±3	local	~9.1GB	2026-10
	more from StartLux
7	StartLux-Decision v1.1 27B StartLux	73±1	model files
29	StartLux-Decision v1.1 9B StartLux	68±1	model files
54	StartLux-Decision v1.1 4B StartLux	65±1	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #126 of 250 · 6.4% of its score · 6 tests

Beats Tern-35B-A3B, GEV-26B-Decide, Cerno 4B v1.1 here, though they rank higher overall. Trails OneJev1.6-0.8B, TokenRhythm NeoHorse-Jev-4B, razorback16 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) submitted	80.1% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 118/236	easy · avg 70%	▇ 2.0%	sharp · noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	94.6% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 122/246	very easy · avg 92%	▇ 1.6%	sharp · maxed out · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	71.6% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 177/243	medium · avg 37%	▇ 1.0%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	90.4% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 222/246	easy · avg 72%	▇ 0.9%	noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	65.9% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 159/246	medium · avg 60%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	78.5% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #34 of 258 · 5.6% of its score · 7 tests

Beats Perplexity Decider v1.1 (27B), H2O Lightning 31B v1.1, AutoTrust-JEV-27B v1.1 here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	40.7% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 105/246	medium · avg 38%	▇ 2.1%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	60.4% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 215/246	medium · avg 47%	▇ 1.9%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	87.8% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 211/246	easy · avg 74%	▇ 0.5%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	89.2% ✎	█▅▅▄▃▃▁▁▁▁·▁ beats 209/245	easy · avg 70%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	90.6% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 136/238	easy · avg 83%	▇ 0.3%	sharp · noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	89.8% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 175/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	51.3% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #130 of 256 · 20% of its score · 9 tests

Beats Jebadiah 9B v2.1 (T1e), Solar Decide, muria (Gemma 4 12B) here, though they rank higher overall. Trails vLLM-SR d3-nano (2B), JevK5 v0.3, Decider 4B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	72.2% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 88/246	easy · avg 74%	▇▇ 4.6%	typical
Decision Index 0.4.1: ANLI (macro-F1) submitted	64.4% ✎	▁▂▃▅▅█▂····▁ beats 125/246	medium · avg 60%	▇▇ 4.5%	at its level
Decision Index 0.4.1: HellaSwag (accuracy) submitted	88.4% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 123/246	very easy · avg 85%	▇▇ 4.4%	sharp
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	47.3% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 121/246	medium · avg 44%	▇ 2.0%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	57.9% ✎	▁··▂▃▃▅▇█▇▃▃ beats 231/246	hard · avg 22%	▇ 1.0%	at its level · noisy
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	84.2% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 212/246	easy · avg 71%	▇ 0.9%	noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	77.6% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 192/246	medium · avg 62%	▇ 0.9%	at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	81.1% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 192/246	easy · avg 66%	▇ 0.7%	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) submitted	93.2% ✎	▁···▁▁▁▁▃▆█▃ beats 193/246	easy · avg 84%	▇ 0.6%	noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #93 of 247 · 8.0% of its score · 6 tests

Beats Jev 1.13, jev-brain · Gemma-4-31B (decision readout), Jeffan-4B-2010 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	68.6% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 140/246	medium · avg 61%	▇ 3.4%	at its level
Decision Index 0.4.1: Humicroedit (accuracy) submitted	63.1% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 176/246	medium · avg 60%	▇ 1.6%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	90.9% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 152/246	easy · avg 83%	▇ 1.0%	sharp · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	67.2% ✎	▁········▁▇█ beats 225/246	medium · avg 60%	▇ 0.9%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	77.7% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 204/237	hard · avg 21%	▇ 0.8%	sharp · at its level · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	61.6% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 216/246	medium · avg 47%	▇ 0.3%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #167 of 263 · 38% of its score · 14 tests

Trails Jevstral 8B, Jobe Qwen3.5-4B (decision readout), Metask-Jev-4B 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	44.4% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 96/246	medium · avg 46%	▇▇▇▇ 8.9%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	66.8% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 83/246	easy · avg 70%	▇▇ 6.0%	sharp
Decision Index 0.4.1: BBH (accuracy) submitted	77.0% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 172/246	easy · avg 66%	▇▇ 5.4%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	88.7% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 87/246	very easy · avg 90%	▇ 3.3%	sharp · maxed out
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	51.9% ✎	▁··▄█▆▆▆▅▅▃▁ beats 142/246	medium · avg 45%	▇ 3.0%	sharp · at its level
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	34.2% ✎	▂▆▅▅▆█▆▃▁··▁ beats 81/246	medium · avg 39%	▇ 2.7%	blunt · at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	78.1% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 119/246	easy · avg 78%	▇ 2.7%	typical
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	95.8% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 86/246	very easy · avg 96%	▇ 1.7%	sharp · maxed out · noisy

6 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	54.8% ✎	▁▁▁▁▁▁▁▂▅█▂▃ beats 240/246	hard · avg 22%	▇ 1.2%	at its level
Decision Index 0.4.1: MuSR (accuracy) submitted	65.0% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 176/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	40.4% ✎	██▄▃▃▄▁▁···▁ beats 60/246	medium · avg 58%	▇ 0.9%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	83.3% ✎	▁███▆▅▃▃▁▁▁▁ beats 192/246	easy · avg 71%	▇ 0.8%	at its level
Decision Index 0.4.1: ChessBench (accuracy) submitted	21.2% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 176/246	hard · avg 16%	▇ 0.6%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	13.8% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 167/246	very hard · avg 13%	▇ 0.0%	blunt

vision▁·▃▂▃▃▇▄▄█▄▃ #50 of 79 * · 1.3% of its score · 1 test

Trails JPT-4B Qwen3.5-4B LoRA, LiquidAI d1-3B, JPT-9B Qwen3.5-9B LoRA here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
JevalKit: VisualDecisionBench images	79.6% ✎	▄▄···█▄·█▄█▄ beats 4/10	easy · avg 78%	▇ 1.3%	blunt

workflows▁▁▁▂▁▂▂▆█▄▁▁ #27 of 114 · 9.0% of its score · 4 tests

Beats d1.1 (hosted), Span-02 (Gemma 4 26B-A4B), Mercury Decide (free endpoint) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
JevalKit: Procedural Typed Decisions	76.5% ✎	█·███▄·▄▄▄·▄ beats 8/12	easy · avg 67%	▇▇ 5.3%	sharp · at its level
JevalKit: DecisionBench standard eval	78.0% ✎	▃▃·▃···▅█▅·█ beats 8/12	easy · avg 70%	▇ 2.3%	at its level
JevalKit: JevBench public 231	85.7% ✎	█▄▄▄▄▄██▄··▄ beats 6/12	easy · avg 82%	▇ 1.3%	typical
JevalKit: Typed Decisions held-out modal labels	78.0% ✎	▃▅··▃·▃▃·█▃█ beats 7/12	easy · avg 74%	▇ 0.1%	blunt

other tests5 not in a task · 11% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	51.0700 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 149/247	avg 41.206 points	▇▇ 4.6%	blunt
JevalKit: JevAlt-Bench EN/TR/DE	90.2% ✎	▅··▃▃▅▃█··▅▃ beats 8/12	very easy · avg 87%	▇ 2.9%	typical
Decision Index 0.4.1: private new domains (equated points)	23.6800 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 81/247	avg 38.861 points	▇ 2.5%	blunt
JevalKit: Decision Index public 0.3 (official weighted points)	59.1800 points ✎	▃▃·▃·▃▃▃▅▅·█ beats 8/12	avg 39.649 points	▇ 1.1%	blunt
JevalKit: Chinese decision validation v0.2	89.3% ✎	▃····█·█▃▃█▃ beats 7/12	very easy · avg 88%	▇ 0.2%	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/startlux-models ↗

$ how the score was worked out · Decision Index 0.4.1 85%, JevalKit (Valen Team) 15%

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	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 85%	CC BY 4.0
JevalKit (Valen Team)	8	▇▇▇▇ 15%	Apache-2.0
prior · pull toward the middle		▇ 1%
weighted average		= 55.3

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	78.5%	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]	51.3%	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