Amber 4B

+ contribute

Amber 4B is an open-weights decision model from MLukauskas. Its published size is 4.21B parameters. Licence: cc-by-nc-4.0.

Catalogue identity notes

Exact Decision Index tested engine amber-4b; reported model Amber 4B. Type: LoRA + head; 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.

55.4±1.3#155 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of Amber 4B, out of 357 with a score, then more from MLukauskas.

#	model	score	where	released
155	muria (Gemma 4 12B) plantwaterer	55±1	model files
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
	more from MLukauskas
145	Amber Hybrid 4B (decision readout) MLukauskas	57±1	model files
274	Amber 0.4B MLukauskas	37±1	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #64 of 250 · 7.4% of its score · 6 tests

Beats OneShot 27B, mogan-decision-31B-it, Sieve 27B v2 here, though they rank higher overall. Trails baselabs joey-1 (2B), Solomon v1.1 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	63.8% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 77/236	easy · avg 70%	▇ 2.3%	sharp · noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	97.2% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 184/246	very easy · avg 92%	▇ 1.9%	sharp · maxed out · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	47.7% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 127/243	medium · avg 37%	▇ 1.2%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	91.2% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 231/246	easy · avg 72%	▇ 1.1%	noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	66.5% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 175/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	77.5% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #139 of 258 · 6.5% of its score · 7 tests

Beats Quyet-1.0-Large, Decider chat · Qwen3.6-27B (decision readout), Xor 1.2 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) submitted	45.1% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 154/246	medium · avg 38%	▇ 2.4%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	51.9% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 118/246	medium · avg 47%	▇ 2.3%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	67.5% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 89/246	easy · avg 74%	▇ 0.6%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	68.1% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 121/245	easy · avg 70%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	88.4% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 105/238	easy · avg 83%	▇ 0.4%	sharp · noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	81.8% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 111/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	64.5% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #73 of 256 · 23% of its score · 9 tests

Beats Turion Jev 31B, AJev lm6q (Gemma 4 31B), oenogarum (Gemma 4 31B) here, though they rank higher overall. Trails Sieve 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	77.8% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 125/246	easy · avg 74%	▇▇ 5.4%	typical
Decision Index 0.4.1: ANLI (macro-F1) submitted	69.8% ✎	▁▂▃▅▅█▂····▁ beats 155/246	medium · avg 60%	▇▇ 5.3%	at its level
Decision Index 0.4.1: HellaSwag (accuracy) submitted	94.4% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 182/246	very easy · avg 85%	▇▇ 5.1%	sharp
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	42.1% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 99/246	medium · avg 44%	▇ 2.3%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	33.0% ✎	▁··▂▃▃▅▇█▇▃▃ beats 177/246	hard · avg 22%	▇ 1.2%	at its level · noisy
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	75.4% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 120/246	easy · avg 71%	▇ 1.1%	noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	72.9% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 153/246	medium · avg 62%	▇ 1.0%	at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	72.9% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 114/246	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) submitted	93.4% ✎	▁···▁▁▁▁▃▆█▃ beats 205/246	easy · avg 84%	▇ 0.7%	noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #111 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) submitted	74.2% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 183/246	medium · avg 61%	▇▇ 4.0%	at its level
Decision Index 0.4.1: Humicroedit (accuracy) submitted	61.7% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 142/246	medium · avg 60%	▇ 1.9%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	89.9% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 137/246	easy · avg 83%	▇ 1.2%	sharp · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	67.3% ✎	▁········▁▇█ beats 228/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	1.9% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 44/237	hard · avg 21%	▇ 0.9%	sharp · at its level · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	56.3% ✎	▁▂▄█▇▅▁▃▄▄▁▁ beats 188/246	medium · avg 47%	▇ 0.3%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #134 of 263 · 45% of its score · 14 tests

Beats Winnow-12B (DecisionBench b6ac22b0), Cerno 4B v1.1, Amber Hybrid 4B (decision readout) here, though they rank higher overall. Trails djev diffgemma (decision readout), Kev 9B v2, Bespoke Nimble 9B v2 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	51.3% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 117/246	medium · avg 46%	▇▇▇▇ 10%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	75.0% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 121/246	easy · avg 70%	▇▇▇ 7.0%	sharp
Decision Index 0.4.1: BBH (accuracy) submitted	70.5% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 135/246	easy · avg 66%	▇▇▇ 6.4%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	94.1% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 127/246	very easy · avg 90%	▇▇ 3.8%	sharp · maxed out
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	61.2% ✎	▁··▄█▆▆▆▅▅▃▁ beats 163/246	medium · avg 45%	▇ 3.5%	sharp · at its level
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	38.3% ✎	▂▆▅▅▆█▆▃▁··▁ beats 104/246	medium · avg 39%	▇ 3.2%	blunt · at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	86.6% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 173/246	easy · avg 78%	▇ 3.1%	typical
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	97.9% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 124/246	very easy · avg 96%	▇ 1.9%	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	35.3% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 208/246	hard · avg 22%	▇ 1.4%	at its level
Decision Index 0.4.1: MuSR (accuracy) submitted	58.0% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 104/246	medium · avg 60%	▇ 1.2%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	53.7% ✎	██▄▃▃▄▁▁···▁ beats 124/246	medium · avg 58%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	67.6% ✎	▁███▆▅▃▃▁▁▁▁ beats 116/246	easy · avg 71%	▇ 1.0%	at its level
Decision Index 0.4.1: ChessBench (accuracy) submitted	9.3% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 51/246	hard · avg 16%	▇ 0.7%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	10.4% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 36/246	very hard · avg 13%	▇ 0.0%	blunt

other tests2 not in a task · 8.3% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	47.6800 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 133/247	avg 41.206 points	▇▇ 5.4%	blunt
Decision Index 0.4.1: private new domains (equated points)	27.1100 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 90/247	avg 38.861 points	▇ 3.0%	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/MLukauskas ↗

$ 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		= 55.4

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	77.5%	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]	64.5%	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