whatmodel.sh / decision / reflex-qwen3.5-4b-lora (DecisionBench 9df1cbbb)

reflex-qwen3.5-4b-lora (DecisionBench 9df1cbbb)

+ contribute

reflex-qwen3.5-4b-lora (DecisionBench 9df1cbbb) is an open-weights decision model from kshetrajna12. Licence: mit.

Catalogue identity notes

DecisionBench v1 comparison: kshetrajna12/reflex-qwen3.5-4b-lora, revision 9df1cbbbc1837ba494530a59d3d9d345688ed250; decision-model; adapter jev-compatible-server-v1; probability source two_order_option_letter_logit_softmax. 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 ↗

60.7±1.8#104 of 355 decision modelsfrom 27 benchmark results across 1 source

$ how it compares

The decision models ranked either side of reflex-qwen3.5-4b-lora (DecisionBench 9df1cbbb), out of 357 with a score, then more from kshetrajna12.

#	model	score	where	RAM	released
103	vLLM-SR d3-flash (9B) vLLM Semantic Router	61±1	local	~18GB
104	glm-5.3-flash (DecisionBench openrout) Z.ai	61±2	comparison only		2026-08-26
105	reflex-qwen3.5-4b-lora (DecisionBench 9df1cbbb) kshetrajna12	61±2	model files
106	reflex Qwen3.8-27B-FP8 (pmscr) (decision readout) kshetrajna12	61±1	source code
107	OneJev1.5-9B OmniJev	61±4	model files		2026-10-13
	more from kshetrajna12
80	reflex-27b (DecisionBench 1d4bf0f2) kshetrajna12	63±2	model files

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

$ what it's good at

workflows▁▁▁▂▁▂▂▆█▄▁▁ #22 of 114 · 99% of its score · 27 tests

Beats Nace.AI Drex Max v1 (27B), d1.1 (hosted), Span-02 (Gemma 4 26B-A4B) 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	65.4% ✎	▂▁▂▃▄▃▁▁▂█▇▄ beats 52/56	medium · avg 51%	▇▇▇▇▇▇▇▇ 21%	at its level
DecisionBench v1 · routing (all-request accuracy) submitted	76.3% ✎	▁▁▁▁▂▂▁▁▁▁▃█ beats 46/56	medium · avg 64%	▇▇▇▇▇ 13%	sharp
DecisionBench v1 · navigation (all-request accuracy) submitted	71.0% ✎	▁▁·▃▁▁▁▁▁▁▃█ beats 38/56	medium · avg 58%	▇▇▇▇ 10%	sharp · at its level
DecisionBench v1 · record classification (all-request accuracy) submitted	47.0% ✎	▃▂▁▂▂▂▄▁▃▆██ beats 44/56	medium · avg 39%	▇▇▇▇ 10%	at its level
DecisionBench v1 · verification (all-request accuracy) submitted	63.0% ✎	▁▁▅▁▃▂▂▁▁▇▇█ beats 49/56	medium · avg 56%	▇▇▇ 6.7%	blunt · at its level
DecisionBench v1 · document record classification (all-request accuracy) submitted	56.2% ✎	▂▂·▁▂▄▃▆▄█▁▁ beats 49/56	medium · avg 46%	▇▇ 5.6%	at its level
DecisionBench v1 · rubric scoring prioritization (all-request accuracy) submitted	47.9% ✎	▁▁▅▅▃▃▅█▃▃·▁ beats 36/56	medium · avg 45%	▇▇ 4.7%	at its level
DecisionBench v1 · triage (all-request accuracy) submitted	87.0% ✎	▁·▁·▂▃▄▃▂█▄▂ beats 45/56	easy · avg 78%	▇ 3.7%	sharp

19 more tests

test	result	vs the others in its run	difficulty	counts	why
DecisionBench v1 · completion assessment (all-request accuracy) submitted	93.0% ✎	▁▁▂▃▄·▂▁▁▆█▂ beats 44/56	very easy · avg 86%	▇ 3.2%	sharp · maxed out
DecisionBench v1 · review scoring (all-request accuracy) submitted	59.0% ✎	▁······▂▃▅▇█ beats 41/56	medium · avg 52%	▇ 3.0%	at its level
DecisionBench v1 · content classification (all-request accuracy) submitted	80.0% ✎	▁····▁▁▁▁▁▁█ beats 45/56	easy · avg 73%	▇ 3.0%	typical
DecisionBench v1 · retrieval ranking (all-request accuracy) submitted	59.0% ✎	▁▁·▁▁▂▃▂▃▁▁█ beats 48/56	medium · avg 35%	▇ 2.6%	at its level · noisy
DecisionBench v1 · target selection (all-request accuracy) submitted	48.0% ✎	▁▁·▁▁▁▃▁▂▁▁█ beats 48/56	hard · avg 29%	▇ 2.5%	at its level · noisy
DecisionBench v1 · bounded extraction (all-request accuracy) submitted	93.7% ✎	▃▃▂·▁▁▁▁▁▁█▅ beats 35/56	very easy · avg 86%	▇ 2.4%	sharp · maxed out · noisy
DecisionBench v1 · risk scoring (all-request accuracy) submitted	66.0% ✎	▁▁·▁▂▂▁▃▃█▇▃ beats 40/56	medium · avg 62%	▇ 2.0%	blunt · at its level
DecisionBench v1 · reasoning (all-request accuracy) submitted	59.3% ✎	▂·▁▁▆█▅▃▃▂▁▁ beats 41/56	medium · avg 52%	▇ 1.3%	at its level
DecisionBench v1 · document workflows (all-request accuracy) submitted	64.6% ✎	█▄▃▃▁▃▂▂▁·▁▁ beats 34/56	easy · avg 66%	▇ 0.8%	blunt · at its level
DecisionBench v1 · retrieval verification (all-request accuracy) submitted	85.5% ✎	█▁▂▁▁▁▁▁▃▅▁▁ beats 37/56	easy · avg 74%	▇ 0.7%	noisy
DecisionBench v1 · guardrails moderation (all-request accuracy) submitted	61.8% ✎	▁▁▁▂▆▃▃█···▁ beats 39/56	medium · avg 52%	▇ 0.6%	blunt · at its level · noisy
DecisionBench v1 · document classification (all-request accuracy) submitted	23.0% ✎	▂▁·▂▁▁▁▄▁▂▃█ beats 31/56	hard · avg 21%	▇ 0.4%	blunt · at its level · noisy
DecisionBench v1 · safety gating (all-request accuracy) submitted	98.0% ✎	▂·····▆█▂▂▄▇ beats 53/56	easy · avg 77%	▇ 0.4%	noisy
DecisionBench v1 · semantic filtering (all-request accuracy) submitted	83.0% ✎	▁·▁▁·▁▁█▂▃█▂ beats 40/56	easy · avg 71%	▇ 0.3%	blunt · noisy
DecisionBench v1 · content moderation (all-request accuracy) submitted	100.0% ✎	▁····▃▅▃▂▂·█ beats 49/56	easy · avg 82%	▇ 0.2%	noisy
DecisionBench v1 · context management (all-request accuracy) submitted	90.5% ✎	▁·····▁▇▆▁█▃ beats 45/56	easy · avg 73%	▇ 0.2%	blunt · noisy
DecisionBench v1 · entity alignment (all-request accuracy) submitted	76.6% ✎	▁·▁▄▇▄▅▄▂▁▄█ beats 31/56	easy · avg 66%	▇ 0.2%	at its level · noisy
DecisionBench v1 · function agent skill routing (all-request accuracy) submitted	74.3% ✎	▃▃▄▅▆▂▅▄▃▄█▃ beats 27/56	easy · avg 68%	▇ 0.1%	blunt · at its level · noisy
DecisionBench v1 · routing triage (all-request accuracy) submitted	0.0% ✎	█▁···▁·▂▆▅▂▂ beats 2/56	medium · avg 42%	▇ 0.0%	blunt · at its level · noisy

$ 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 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/kshetrajna12 ↗

$ 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		= 60.7

$ 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