whatmodel.sh / decision / bge-reranker-v2-m3 (DecisionBench 953dc6f6)

bge-reranker-v2-m3 (DecisionBench 953dc6f6)

+ contribute

bge-reranker-v2-m3 (DecisionBench 953dc6f6) is an open-weights decision model from BAAI. Its published size is 0.567755777B parameters. Licence: apache-2.0.

Catalogue identity notes

DecisionBench v1 comparison: BAAI/bge-reranker-v2-m3, revision 953dc6f6f85a1b2dbfca4c34a2796e7dde08d41e; classifier; adapter jev-compatible-server-v1; probability source native_probability_distribution. Kept separate from newer or unpinned Decision Index recipes; standard prompt, not compact. No unverified runtime or OpenRouter price imported. 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 ↗

Published served parameter count; not a RAM or download-size measurement.

18.8±1.8#346 of 355 decision modelsfrom 27 benchmark results across 1 source

$ how it compares

The decision models ranked either side of bge-reranker-v2-m3 (DecisionBench 953dc6f6), out of 357 with a score.

#	model	score	where	released
346	NanoJev (DecisionBench 4a19595e) C-Tianyu	19±2	model files
347	Bekko System One v0 17M hotchpotch	19±1	model files
348	bge-reranker-v2-m3 (DecisionBench 953dc6f6) BAAI	19±2	model files
349	system-one-gemma akash-kamat	18±1	source code
350	Supersonic Labs Julia 1 Supersonic Labs	18±1	model files	2026-09-25

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

$ what it's good at

workflows▁▃▁▂▁▂▂▆█▄▁▁ #108 of 114 · 99% of its score · 27 tests

Beats CLM-v0.1-8B Qwen3-8B heads here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
DecisionBench v1 · action selection (all-request accuracy) submitted	22.3% ✎	▂▁▃▃▄▃▁▁▂█▇▄ beats 6/56	medium · avg 51%	▇▇▇▇▇▇ 15%	typical
DecisionBench v1 · completion assessment (all-request accuracy) submitted	34.0% ✎	▁▁▃▃▄·▂▁▁▆█▂ beats 4/56	very easy · avg 86%	▇▇▇▇▇ 12%	sharp · at its level
DecisionBench v1 · routing (all-request accuracy) submitted	22.7% ✎	▁▁▁▃▂▂▁▁▁▁▃█ beats 7/56	medium · avg 64%	▇▇▇▇▇ 12%	sharp
DecisionBench v1 · bounded extraction (all-request accuracy) submitted	27.2% ✎	▃▃▂·▁▁▁▁▁▁█▅ beats 11/56	very easy · avg 86%	▇▇▇ 8.4%	sharp · noisy
DecisionBench v1 · triage (all-request accuracy) submitted	32.0% ✎	▁·▁·▃▃▄▃▂█▄▂ beats 3/56	easy · avg 78%	▇▇▇ 7.6%	sharp · at its level
DecisionBench v1 · navigation (all-request accuracy) submitted	9.0% ✎	▁▃·▃▁▁▁▁▁▁▃█ beats 2/56	medium · avg 58%	▇▇▇ 6.9%	sharp
DecisionBench v1 · verification (all-request accuracy) submitted	31.5% ✎	▃▁▅▁▃▂▂▁▁▇▇█ beats 1/56	medium · avg 56%	▇▇▇ 6.6%	blunt · at its level
DecisionBench v1 · record classification (all-request accuracy) submitted	15.0% ✎	▃▃▁▂▂▂▄▁▃▆██ beats 6/56	medium · avg 39%	▇▇ 5.9%	typical

19 more tests

test	result	vs the others in its run	difficulty	counts	why
DecisionBench v1 · content classification (all-request accuracy) submitted	18.0% ✎	▁····▁▃▁▁▁▁█ beats 4/56	easy · avg 73%	▇▇ 4.6%	at its level
DecisionBench v1 · document record classification (all-request accuracy) submitted	25.2% ✎	▂▂·▃▂▄▃▆▄█▁▁ beats 7/56	medium · avg 46%	▇▇ 4.0%	typical
DecisionBench v1 · rubric scoring prioritization (all-request accuracy) submitted	26.1% ✎	▁▁▅▅▃▃▅█▃▃·▁ beats 11/56	medium · avg 45%	▇ 3.4%	typical
DecisionBench v1 · review scoring (all-request accuracy) submitted	32.0% ✎	▁······▂▃▅▇█ beats 10/56	medium · avg 52%	▇ 2.6%	at its level
DecisionBench v1 · risk scoring (all-request accuracy) submitted	37.0% ✎	▁▁·▁▃▂▁▃▃█▇▃ beats 6/56	medium · avg 62%	▇ 2.2%	blunt · at its level
DecisionBench v1 · retrieval verification (all-request accuracy) submitted	51.3% ✎	█▁▂▁▁▁▁▁▂▅▁▁ beats 16/56	easy · avg 74%	▇ 1.2%	at its level · noisy
DecisionBench v1 · reasoning (all-request accuracy) submitted	30.4% ✎	▂·▁▁▆█▅▃▃▂▁▁ beats 6/56	medium · avg 52%	▇ 1.1%	at its level
DecisionBench v1 · document workflows (all-request accuracy) submitted	51.4% ✎	█▄▃▁▁▃▂▂▁·▁▁ beats 22/56	easy · avg 66%	▇ 1.0%	blunt · at its level
DecisionBench v1 · retrieval ranking (all-request accuracy) submitted	13.0% ✎	▁▁·▁▁▂▃▃▃▁▁█ beats 19/56	medium · avg 35%	▇ 0.8%	noisy
DecisionBench v1 · target selection (all-request accuracy) submitted	8.0% ✎	▁▁·▁▁▁▃▁▂▁▁█ beats 13/56	hard · avg 29%	▇ 0.7%	too hard for it · noisy
DecisionBench v1 · safety gating (all-request accuracy) submitted	65.0% ✎	▂·····▆█▂▂▄▇ beats 27/56	easy · avg 77%	▇ 0.7%	at its level · noisy
DecisionBench v1 · guardrails moderation (all-request accuracy) submitted	51.1% ✎	▁▁▁▂▆▃▃█···▁ beats 29/56	medium · avg 52%	▇ 0.5%	blunt · at its level · noisy
DecisionBench v1 · semantic filtering (all-request accuracy) submitted	63.0% ✎	▁·▁▁·▁▁█▃▃█▂ beats 24/56	easy · avg 71%	▇ 0.4%	blunt · at its level · noisy
DecisionBench v1 · content moderation (all-request accuracy) submitted	46.0% ✎	▁····▃▅▃▂▂·█ beats 8/56	easy · avg 82%	▇ 0.4%	at its level · noisy
DecisionBench v1 · context management (all-request accuracy) submitted	52.0% ✎	▁·····▁▇▆▁█▃ beats 11/56	easy · avg 73%	▇ 0.3%	blunt · at its level · noisy
DecisionBench v1 · document classification (all-request accuracy) submitted	5.0% ✎	▂▁·▂▁▃▁▄▁▂▂█ beats 12/56	hard · avg 21%	▇ 0.2%	blunt · noisy
DecisionBench v1 · entity alignment (all-request accuracy) submitted	98.1% ✎	▁·▁▄▇▄▅▄▂▁▄█ beats 43/56	easy · avg 66%	▇ 0.2%	at its level · noisy
DecisionBench v1 · function agent skill routing (all-request accuracy) submitted	86.4% ✎	▃▃▄▅▆▂▅▄▃▄█▃ beats 34/56	easy · avg 68%	▇ 0.1%	blunt · at its level · noisy
DecisionBench v1 · routing triage (all-request accuracy) submitted	59.5% ✎	█▁···▁·▂▆▅▂▂ beats 37/56	medium · avg 42%	▇ 0.0%	blunt · at its level · noisy

$ where to run it

Download model files ↗
Published model files verified. This is the revision recorded by the benchmark. 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/BAAI ↗

$ how the score was worked out · Hanno Labs DecisionBench v1 r… 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
Hanno Labs DecisionBench v1 reviewed results · submitted; independent execution unverified	27	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%	CC0 1.0
prior · pull toward the middle		▇ 1%
weighted average		= 18.8

$ 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