Sieve 2B

+ contribute

Sieve 2B is an open-weights decision model from Sthānika AI, released 13 October 2026. Its published size is 2.27B parameters. Licence: apache-2.0.

Catalogue identity notes

Exact Decision Index tested engine sieve-2b; reported model Sieve 2B. 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

Release date ↗ · Weights licence ↗ · Parameter count ↗

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

36.5±1.1#279 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of Sieve 2B, out of 357 with a score, then more from Sthānika AI.

#	model	score	where	RAM	released
279	Crivo-400M tardellirs	37±1	model files
280	InternLM Intern-Decision Qwen3.5-2B Shanghai AI Laboratory (InternLM)	37±1	model files
281	Sieve 2B Sthānika AI	37±1	model files		2026-10-13
282	smalljev-semantic-v9 (DecisionBench 8e2bac7c) isHeSatoshi	36±2	model files
283	Strands Decider 2B (Hobson v21) Amazon Web Services	36±1	local	~6.6GB	2026-10
	more from Sthānika AI
3	Sieve 31B v3 Sthānika AI	74±2	model files		2026-10-27
37	Sieve 27B v2 Sthānika AI	67±1	model files		2026-10-18
79	Sieve-Vision-27B Sthānika AI	63±4	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▃▃▄▆██▆ #206 of 250 · 9.4% of its score · 6 tests

Beats Cam1, mini-jev Qwen3-4B (decision readout), InternLM Intern-Decision Qwen3.5-2B here, though they rank higher overall. Trails JevK5 Lite, Mola-68M, Crivo-68M here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	78.1% ✎	▁▂▁▁▁▃▂▄▅▆█▁ beats 43/246	very easy · avg 92%	▇▇ 4.9%	sharp · noisy
Decision Index 0.4.1: API-Bank (accuracy) submitted	45.7% ✎	▁▁▁▁▁▁▁▃▃▂█▂ beats 38/236	easy · avg 70%	▇ 2.3%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	18.9% ✎	▁▃▁▂▃▅▅▆█▇█▃ beats 5/246	easy · avg 72%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	57.8% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 53/246	medium · avg 60%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	1.1% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 52/243	medium · avg 37%	▇ 0.4%	sharp · noisy
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	79.6% ✎	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▁▁▁·▁▃▁▁▃▃█▅ #240 of 258 · 4.9% of its score · 7 tests

Trails Prumo (rule-based engine), MoganAI Lavoir ModernBERT-large, Bekko System One v0 68M here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	24.5% ✎	▁▁▁▁▁▁▃▂▃▃█▅ beats 18/246	medium · avg 47%	▇ 1.5%	at its level
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	33.4% ✎	▁▁▁▁▁▁▁▂▃▄▇█ beats 63/246	medium · avg 38%	▇ 1.5%	blunt · at its level
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	51.3% ✎	▁▁▁▁▁▁▁▃▃▆█▄ beats 27/238	easy · avg 83%	▇ 0.6%	sharp · at its level · noisy
Decision Index 0.4.1: HoVer (accuracy) submitted	50.9% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 11/246	easy · avg 74%	▇ 0.5%	blunt · at its level
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	50.0% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 21/245	easy · avg 70%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	75.3% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 66/238	easy · avg 79%	▇ 0.3%	at its level · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	38.9% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #208 of 256 · 24% of its score · 9 tests

Beats Valen-2B, Saina Helm 2 Omni (0.8B), Gevva E2B here, though they rank higher overall. Trails DecisionTune 1.0, vera-spark, JEV FLAN-T5 Large public v2 (342M) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: HellaSwag (accuracy) submitted	62.6% ✎	▂▄▂▃▄▇▃▇█▅▃▁ beats 48/246	very easy · avg 85%	▇▇▇▇ 9.3%	sharp · at its level
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	65.4% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 59/246	easy · avg 74%	▇▇ 5.3%	at its level
Decision Index 0.4.1: ANLI (macro-F1) submitted	40.8% ✎	▁▂▃▅▅█▂····▁ beats 38/246	medium · avg 60%	▇▇ 4.2%	at its level
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	28.5% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 53/246	medium · avg 44%	▇ 1.4%	at its level
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	39.3% ✎	▁▁▁▁▁▃▁▂▂▇▇█ beats 23/246	easy · avg 71%	▇ 1.0%	at its level · noisy
Decision Index 0.4.1: FinEntity (macro-F1) submitted	68.1% ✎	▁···▁▁▁▁▃▆█▃ beats 39/246	easy · avg 84%	▇ 0.9%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	41.5% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 37/246	medium · avg 62%	▇ 0.8%	at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	7.0% ✎	▁▁▁▁▃▁▁▁▁▃▆█ beats 10/246	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: ACOS (per-review F1) submitted	6.3% ✎	▁··▂▃▃▅▇█▇▃▃ beats 33/246	hard · avg 22%	▇ 0.4%	noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #181 of 247 · 7.5% of its score · 6 tests

Beats Sieve 9B, Kev 9B v2, vLLM-SR d3-nano (2B) here, though they rank higher overall. Trails Bekko System One v0 68M, Crivo-17M, Crivo-32M v2 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	43.9% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 45/246	medium · avg 61%	▇ 3.2%	at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	70.1% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 66/246	easy · avg 83%	▇ 1.7%	sharp · at its level · noisy
Decision Index 0.4.1: Humicroedit (accuracy) submitted	57.5% ✎	▁·▁▃▂▃▅█▄▂▁▁ beats 70/246	medium · avg 60%	▇ 1.4%	blunt · at its level
Decision Index 0.4.1: cfcolor (accuracy) submitted	58.7% ✎	▁········▁▇█ beats 90/246	medium · avg 60%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	42.8% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 107/246	medium · avg 47%	▇ 0.2%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	1.1% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 17/237	hard · avg 21%	▇ 0.2%	sharp · too hard for it · noisy

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #186 of 263 · 48% of its score · 14 tests

Beats baselabs joey-1 (2B), Aurelis-2B, StartLux-Decision v1.1 0.8B here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	83.5% ✎	▂▂▂▂▂▂▃▃▄▆█▆ beats 74/246	very easy · avg 90%	▇▇▇▇ 10%	sharp · at its level
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	93.4% ✎	▁▁▂▁▂▂▂▃▃▆█▅ beats 75/246	very easy · avg 96%	▇▇▇ 8.5%	sharp · noisy
Decision Index 0.4.1: MMLU (accuracy) submitted	60.3% ✎	▁▄▄▃▃▃▅▇▅█▇▁ beats 75/246	easy · avg 70%	▇▇▇ 7.0%	sharp · at its level
Decision Index 0.4.1: MMLU-Pro (accuracy) submitted	35.2% ✎	▄▄▃▃▃▄▅▇██▅▁ beats 73/246	medium · avg 46%	▇▇ 6.0%	sharp · at its level
Decision Index 0.4.1: BBH (accuracy) submitted	46.5% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 55/246	easy · avg 66%	▇▇ 5.6%	at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	55.4% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 32/246	easy · avg 78%	▇ 3.5%	at its level
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	29.6% ✎	▂▆▅▅▆█▆▃▁··▁ beats 56/246	medium · avg 39%	▇ 2.0%	blunt · at its level
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	21.3% ✎	▁··▄█▆▆▆▅▅▃▁ beats 45/246	medium · avg 45%	▇ 1.9%	sharp · 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	51.6% ✎	▁▁▃█▅▄▁▁▁▁▁▁ beats 41/246	medium · avg 60%	▇ 0.9%	blunt · at its level
Decision Index 0.4.1: CLadder (accuracy) submitted	54.5% ✎	▁███▆▅▃▃▁▁▁▁ beats 40/246	easy · avg 71%	▇ 0.9%	at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	30.5% ✎	██▄▃▃▄▁▁···▁ beats 3/246	medium · avg 58%	▇ 0.8%	at its level · noisy
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	8.5% ✎	▁▁▁▁▁▁▃▂▅█▂▁ beats 40/246	hard · avg 22%	▇ 0.6%	typical
Decision Index 0.4.1: ChessBench (accuracy) submitted	9.4% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 53/246	hard · avg 16%	▇ 0.3%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	12.2% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 118/246	very hard · avg 13%	▇ 0.0%	blunt

other tests2 not in a task · 5.9% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	22.8800 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 46/247	avg 41.206 points	▇▇ 3.8%	blunt
Decision Index 0.4.1: private new domains (equated points)	20.4400 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 76/247	avg 38.861 points	▇ 2.1%	blunt

$ where to run it

Download model files ↗
Published model files verified. Includes adapters or readout weights; follow the model documentation for the required base model. 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/sthanika-ai ↗

$ how the score was worked out · Decision Index 0.4.1 100%

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	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 100%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 36.5

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	79.6%	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]	38.9%	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