Saif 4B

+ contribute

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

Catalogue identity notes

Exact Decision Index tested engine saif-4b; reported model Saif 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.

53.6±1.2#177 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of Saif 4B, out of 357 with a score.

#	model	score	where	released
177	JevAny-Gemma-4B (pointer) (JevBench mini) SimpleJev	54±2	source code
178	JPT-9B Qwen3.5-9B LoRA kirp	54±1	model files
179	Saif 4B Gyr Technology	54±1	model files
180	SemIf (DecisionBench 1f2dea3e) TheoLeeCJ	54±2	source code
181	RSI-Jev v6.1 VL 4B shgao	53±1	model files	2026-10-07

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #88 of 250 · 7.6% of its score · 6 tests

Beats Sieve 27B v2, Jev 1.13, Xor 26B-A4B v1.2 here, though they rank higher overall. Trails OneJev1.6-0.8B, Jev-Omni, baselabs joey-1 (2B) 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	65.9% ✎	▁▁▁▁▁▁▁▂▃▃█▂ beats 82/236	easy · avg 70%	▇ 2.4%	sharp · at its level · noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	96.3% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 163/246	very easy · avg 92%	▇ 2.1%	sharp · maxed out · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	44.3% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 121/243	medium · avg 37%	▇ 1.1%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	89.6% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 211/246	easy · avg 72%	▇ 1.1%	noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	67.4% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 190/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	79.3% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #134 of 258 · 6.4% of its score · 7 tests

Beats Quyet-1.0-Large, OneJev1.6-4B, Decider chat · Qwen3.6-27B (decision readout) 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	43.3% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 132/246	medium · avg 38%	▇ 2.3%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	52.2% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 123/246	medium · avg 47%	▇ 2.2%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	83.5% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 172/246	easy · avg 74%	▇ 0.6%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	69.0% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 126/245	easy · avg 70%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	94.6% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 201/238	easy · avg 83%	▇ 0.4%	sharp · noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	89.6% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 173/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	63.7% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #117 of 256 · 23% of its score · 9 tests

Beats garum (Gemma 4 31B), Shisa DE-2.1 (repeat), GPT-6 Luna Decisions here, though they rank higher overall. Trails Decider 4B, jiwo 4B v2 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	91.7% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 137/246	very easy · avg 85%	▇▇ 5.6%	sharp
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	78.6% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 140/246	easy · avg 74%	▇▇ 5.4%	typical
Decision Index 0.4.1: ANLI (macro-F1) submitted	63.1% ✎	▁▂▃▅▅█▂····▁ beats 115/246	medium · avg 60%	▇▇ 5.2%	at its level
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	34.4% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 70/246	medium · avg 44%	▇ 2.2%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	31.4% ✎	▁··▂▃▃▅▇█▇▃▃ beats 168/246	hard · avg 22%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	81.6% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 178/246	easy · avg 71%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	58.5% ✎	▁▁▁▂▅▄▄▃▄▆█▄ beats 102/246	medium · avg 62%	▇ 1.0%	at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	74.5% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 127/246	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	92.7% ✎	▁···▁▁▁▁▃▆█▃ beats 184/246	easy · avg 84%	▇ 0.7%	noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #161 of 247 · 9.2% of its score · 6 tests

Beats Jegue 0.1 (9B), Standard One 8B SH (v4) here, though they rank higher overall. Trails Crivo-17M, Crivo-32M v2, Crivo-68M 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	68.8% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 143/246	medium · avg 61%	▇▇ 4.0%	at its level
Decision Index 0.4.1: Humicroedit (accuracy) submitted	63.6% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 184/246	medium · avg 60%	▇ 1.9%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	80.2% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 86/246	easy · avg 83%	▇ 1.2%	sharp · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	61.5% ✎	▁········▁▇█ beats 119/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	13.1% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 111/237	hard · avg 21%	▇ 0.8%	sharp · at its level · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	47.8% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 166/246	medium · avg 47%	▇ 0.3%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #143 of 263 · 45% of its score · 14 tests

Beats Winnow-12B (DecisionBench b6ac22b0), Amber Hybrid 4B (decision readout), StartLux-Decision v1.1 2B here, though they rank higher overall. Trails razorback16 diffgemma (decision readout), Solar Decide Flash, 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: MMLU-Pro (accuracy) submitted	55.7% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 139/246	medium · avg 46%	▇▇▇▇ 10%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	72.4% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 105/246	easy · avg 70%	▇▇▇ 7.1%	sharp · at its level
Decision Index 0.4.1: BBH (accuracy) submitted	65.2% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 105/246	easy · avg 66%	▇▇▇ 6.4%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	93.7% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 119/246	very easy · avg 90%	▇▇ 4.3%	sharp · maxed out
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	42.0% ✎	▁··▄█▆▆▆▅▅▃▁ beats 119/246	medium · avg 45%	▇ 3.4%	sharp · at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	79.4% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 124/246	easy · avg 78%	▇ 3.2%	typical
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	42.4% ✎	▂▆▅▅▆█▆▃▁··▁ beats 127/246	medium · avg 39%	▇ 3.0%	blunt · at its level
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	97.8% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 117/246	very easy · avg 96%	▇ 2.3%	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	31.9% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 169/246	hard · avg 22%	▇ 1.3%	at its level
Decision Index 0.4.1: MuSR (accuracy) submitted	58.1% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 109/246	medium · avg 60%	▇ 1.2%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	47.4% ✎	██▄▃▃▄▁▁···▁ beats 99/246	medium · avg 58%	▇ 1.1%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	63.6% ✎	▁███▆▅▃▃▁▁▁▁ beats 87/246	easy · avg 71%	▇ 1.0%	at its level
Decision Index 0.4.1: ChessBench (accuracy) submitted	11.0% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 74/246	hard · avg 16%	▇ 0.6%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	9.8% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 23/246	very hard · avg 13%	▇ 0.0%	blunt

other tests2 not in a task · 8.0% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	43.7000 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 119/247	avg 41.206 points	▇▇ 5.2%	blunt
Decision Index 0.4.1: private new domains (equated points)	40.6200 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 130/247	avg 38.861 points	▇ 2.8%	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/GyrTechnology ↗

$ 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		= 53.6

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	79.3%	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]	63.7%	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