whatmodel.sh / decision / Solar Decide Flash

Solar Decide Flash

+ contribute

Solar Decide Flash is a closed-weights decision model from Upstage. Licence: proprietary.

Catalogue identity notes

Named native decision endpoint verified 2026-10-31. Medical study records exact requested and returned model IDs and pinned Upstage routing. Historical dated service snapshot retained in each result; no claim that current served weights are immutable.

48.0±1.2#221 of 355 decision modelsfrom 33 benchmark results across 4 sources

$ how it compares

The decision models ranked either side of Solar Decide Flash, out of 357 with a score, then more from Upstage.

#	model	score	where	released
221	mmastrac diffgemma (decision readout) vllm-project	48±1	source code
222	InternLM Intern-Decision Qwen3.5-4B Shanghai AI Laboratory (InternLM)	48±1	model files
223	Solar Decide Flash Upstage	48±1	api price unknown
224	razorback16 diffgemma (decision readout) vllm-project	48±1	source code
225	Interfaze lev Qwen3.5-4B LoRA Interfaze	48±1	model files
	more from Upstage
146	Solar Decide Upstage	57±1	api price unknown	2026-09-22

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

$ what it's good at

retrieval & classification▁▁▁·▁▁▁▃▃▃█▅ #203 of 258 · 16% of its score · 8 tests

Trails Bosun v3.1 0.6B LoRA, Mola-68M, Bosun v3.1 1.7B LoRA here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
System One Bench: MASSIVE intent (Arabic)	63.0% ✎	▄·▃·▁▃▃▃█▄▇▇ beats 3/29	easy · avg 76%	▇▇ 3.9%	at its level
System One Bench: MIRACL pairwise (Arabic)	83.0% ✎	▂···▁··▃▄▂█▇ beats 3/29	easy · avg 84%	▇▇ 3.9%	at its level
LeanMCP · scifact rank noul	66.0% ✎	▂····▃···▃▅█ beats 1/11	medium · avg 62%	▇ 2.7%	at its level
System One Bench: MIRACL rerank (Arabic)	71.5% ✎	▁▂·▁▃·▂·▄▄█▇ beats 4/29	easy · avg 79%	▇ 2.4%	at its level
LeanMCP · nfcorpus rank noul	33.0% ✎	▃▃·····▃▃▅██ beats 1/11	hard · avg 29%	▇ 1.6%	blunt · at its level
System One Bench: MASSIVE scenario (Arabic)	46.0% ✎	▃▃·▄▃▂█▂▁▁·▃ beats 3/29	medium · avg 60%	▇ 1.3%	at its level
System One Bench: AG News (English)	86.0% ✎	▁···▅▁▅▂▆█▂▅ beats 3/29	easy · avg 88%	▇ 0.3%	blunt
System One Bench: Banking77 (English)	1.0% ✎	█▅▆█▅▃▃▃···▂ beats 0/26	very easy · avg 87%	—	failed run: All requests failed; not a measurement of ability. Source records unsupported option cardinality.

language understanding▂▂▂▃▃▅▆▃█▆▄▃ #197 of 256 · 10% of its score · 9 tests

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
System One Bench: OSACT4 offensive (Arabic)	84.5% ✎	▁······▁▃█▇▄ beats 6/29	easy · avg 84%	▇ 2.8%	sharp · at its level
System One Bench: BoolQ (English)	86.0% ✎	▁▁▁▃·▁▃▃▇▇█▅ beats 7/29	easy · avg 86%	▇ 2.6%	at its level
System One Bench: SST-2 (English)	90.5% ✎	▁·▃▁▃▂▃▅█·▂▁ beats 6/29	easy · avg 92%	▇ 1.8%	typical
System One Bench: SST-5 (English)	44.5% ✎	▃·▁▃▃·▇▄▄▄█▅ beats 3/29	medium · avg 51%	▇ 1.0%	blunt · at its level
System One Bench: XNLI Arabic (Arabic)	52.0% ✎	▄▁▃▁▄▇█▅▃▁·▃ beats 0/29	medium · avg 68%	▇ 0.6%	at its level
System One Bench: XNLI French (French)	61.3% ✎	▃▅·▁▁▅▅█▃▁·▃ beats 0/29	medium · avg 76%	▇ 0.6%	blunt · at its level
System One Bench: Tweet sentiment (Arabic)	65.5% ✎	▁······▁▁▄█▄ beats 11/29	medium · avg 62%	▇ 0.4%	blunt · at its level
System One Bench: XNLI English (English)	65.3% ✎	▃·▁·▃▇█▄█▃▃▁ beats 0/29	easy · avg 85%	▇ 0.2%	blunt · noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
System One Bench: DAIR Emotion (English)	54.5% ✎	██▃·····▁··▁ beats 3/29	medium · avg 59%	▇ 0.0%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #139 of 263 · 47% of its score · 7 tests

Beats Winnow-12B (DecisionBench b6ac22b0), Amber Hybrid 4B (decision readout), StartLux-Decision v1.1 2B here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Etsabary · 1342 text decisions	40.1% ✎	▃··▂·▃▅▁▃▂▂█ beats 0/24	medium · avg 47%	▇▇▇▇▇▇ 16%	sharp · at its level
LeanMCP · scienceqa text	85.6% ✎	▄··▄█▄█▄·█▄▄ beats 0/11	very easy · avg 88%	▇▇▇▇ 9.4%	sharp
LeanMCP · medqa usmle	72.2% ✎	▄▄▄·▄▄▄▄▄·██ beats 0/11	easy · avg 72%	▇▇▇ 6.6%	at its level
MedQA · Freyer et al. (first-valid accuracy)	72.9% ✎	█··▃▃·▃▃▃·▃█ beats 0/11	easy · avg 72%	▇▇ 5.1%	at its level
LeanMCP · medmcqa	63.4% ✎	▃·▃··▃█▃·▅▅▃ beats 0/11	medium · avg 63%	▇▇ 3.8%	at its level
LeanMCP · pubmedqa	52.4% ✎	▃···▂·▂·▄▄▂█ beats 0/11	medium · avg 54%	▇ 3.6%	at its level
LeanMCP · mmlu pro	47.2% ✎	▄··▄▄████··▄ beats 0/11	medium · avg 54%	▇ 2.7%	at its level

workflows▁▁▁▂▁▂▂▆█▄▁▁ #68 of 114 · 27% of its score · 10 tests

Beats Clef-omni, baselabs joey-1 (2B), ZTC-Judge-9B (DecisionBench 05d154a5) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
LeanMCP · mcpmark tool selection	94.4% ✎	▃··········█ beats 0/11	very easy · avg 96%	▇▇▇ 7.5%	sharp · maxed out · noisy
LeanMCP · rewardbench2	63.2% ✎	▄▄▄·▄▄▄▄█·█▄ beats 2/11	medium · avg 58%	▇ 3.3%	at its level
System One Bench: Typed decisions (English)	62.5% ✎	▁▁·▁▁▁▅▄▇██▁ beats 5/29	medium · avg 63%	▇ 3.0%	blunt · at its level
LeanMCP · aegis2 response	72.0% ✎	█··▃·▃·█··▅▅ beats 0/11	easy · avg 70%	▇ 2.9%	at its level
LeanMCP · mcpmark server routing	40.5% ✎	▂··▃··▄█▂·▂▄ beats 1/11	medium · avg 36%	▇ 2.7%	at its level
LeanMCP · atbench500	50.0% ✎	▃··▅▅█··▅··▅ beats 0/11	medium · avg 63%	▇ 2.3%	sharp · at its level · noisy
LeanMCP · helpsteer2 helpfulness	1.0920 ordinal levels ✎	▄▄▄▄▄·▄·█▄▄▄ beats 2/10	avg 1.357 ordinal levels	▇ 1.7%	blunt
LeanMCP · sst5	0.7212 ordinal levels ✎	▃·····▂▂▄█▄▂ beats 0/11	avg 0.879 ordinal levels	▇ 1.7%	blunt

2 more tests

test	result	vs the others in its run	difficulty	counts	why
LeanMCP · prompt injections	79.3% ✎	▂▂▂▂█·▄▂···▂ beats 8/11	medium · avg 62%	▇ 1.5%	at its level
LeanMCP · jailbreak classification	97.0% ✎	▂·▄··▂···█▆▂ beats 8/11	very easy · avg 95%	▇ 0.0%	blunt · maxed out · noisy

$ where to run it

Through an API

api · OpenRouter · price unknown · openrouter.ai ↗

Model repository and documentation: openrouter.ai ↗

$ how the score was worked out · LeanMCP · JevBench 54%, System One Bench (Ali Alhashi… 25%, Etsabary · decision benchmarks 16% and 1 more

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
LeanMCP · JevBench	16	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 54%	MIT
System One Bench (Ali Alhashimi)	15	▇▇▇▇▇▇▇ 25%	CC-BY-4.0
Etsabary · decision benchmarks	1	▇▇▇▇▇ 16%	MIT
Freyer et al. · MedQA	1	▇▇ 5%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 48.0

not counted (1)

System One Bench (Ali Alhashimi)

System One Bench: Banking77 (English) [correct]

1.0%

failed run: All requests failed; not a measurement of ability. Source records unsupported option cardinality.

$ 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