whatmodel.sh / Qwen3.5 27B

Qwen3.5 27B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Qwen3.5 27B is an open-weights language model from Alibaba, released February 2026. Its published size is 27.8B parameters and a 262k-token context window. Licence: apache-2.0.

68.5±0.4#101 of 719 LLMsfrom 139 benchmark results across 14 sources

$ how it compares

The LLMs ranked either side of Qwen3.5 27B, out of 828 with a score, then more from Alibaba.

#modelscorewhereRAMreleased
109GPT-5.4 nano OpenAI69±1api $0.46/M2026-03
110Claude Sonnet 4.5 Anthropic69±1api $6.00/M2025-09
111Qwen3.5 27B Alibaba69±0api $0.84/M +local~18GB2026-02
112GLM-5V-Turbo Z.ai69±2api $1.90/M2026-04
113GPT-5.2-Codex OpenAI69±2comparison only2026-01
more from Alibaba
18Qwen3.8 Max (2.4T-A95B) Alibaba78±1api $3.00/M +local~380GB2026-08
34Qwen3.8 Max (0902) Alibaba76±1api $3.00/M2026-09
37Qwen3.8-Flash-Next Alibaba75±1local~59GB2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #85 of 300 · 25% of its score · 18 tests

Beats Grok 4.20 (0309 reasoning), GPT-5.1, Grok 4.3 here, though they rank higher overall. Trails Hy3 preview (295B-A21B), MiniMax-M2.7, Step 3.5 Flash (196B-A11B) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
SWE-bench Verified +468.6% ✎▄▄·····▄▄▄▄█ beats 6/7hard · avg 17%▇▇▇ 7.2%sharp · at its level · several runs
SWE-bench Multilingual +257.7% ✎▄·▄·····█▄█▄ beats 6/7hard · avg 18%▇ 2.5%at its level · several runs
Terminal-Bench 2.141.6% ✎▄······█▄▄█▄ beats 7/7very hard · avg 7.1%▇ 2.0%at its level
τ²-bench aggregate lab +179.0% ✎▁▁▁·▂▂▄▃▃▅█▂ beats 20/37medium · avg 36%▇ 1.7%several runs
SWE-bench Pro (split unspecified)42.1% ✎▄···▄···▄▄██ beats 6/7very hard · avg 6.5%▇ 1.6%sharp · at its level
ContractBench success rate (33 observation-contract tool tasks)64.6% ✎▆··▆▄▂▄▂▆▆█▄ beats 18/23very hard · avg 14%▇ 1.4%sharp · at its level
SWE-bench Pro public set, Qwen-refined lab +151.2% ✎▄······▄█·▄▄ beats 3/5hard · avg 18%▇ 1.3%at its level · several runs
Terminal-Bench 2.0 lab41.6% ✎▂▂▅▅████▆▂▃▃ beats 15/36very hard · avg 11%▇ 1.1%at its level · small run
10 more tests
testresultvs the others in its rundifficultycountswhy
WideSearch lab +161.1% ✎█·▂▆··▂···▂▂ beats 5/10hard · avg 33%▇ 0.9%at its level · small run · several runs
BrowseComp-ZH lab62.1% ✎▁▁▁▁▁▃·▁▄▅▄█ beats 16/23hard · avg 23%▇ 0.9%at its level · small run
OSWorld-Verified lab56.2% ✎█▄·▂▆▂▄▂▆▄▆█ beats 7/25hard · avg 21%▇ 0.8%at its level · small run
BrowseComp lab61.0% ✎▂·▄▁▂▃▆▄█▂▃▄ beats 25/42hard · avg 23%▇ 0.7%at its level · small run
MCP-Atlas (public set) lab68.4% ✎▂·▁▂▁·▁▃█▆▆▂ beats 22/37hard · avg 29%▇ 0.7%at its level · small run
MCPMark lab36.3% ✎▃·····█··▃█▃ beats 6/8very hard · avg 10%▇ 0.6%at its level · small run
Toolathlon (Tool Decathlon) lab31.5% ✎▂▂▄▂▄▄▆▂█▆██ beats 5/27very hard · avg 11%▇ 0.6%at its level · small run
Humanity's Last Exam with tools lab48.5% ✎▂▂▂▁▄▂▃▆█▇▂▆ beats 29/54hard · avg 18%▇ 0.5%at its level · small run
BFCL V4 (December 2025 archive) lab68.5% ✎▁▁▁▂▇▆▆▆█▂▄▂ beats 103/111medium · avg 41%▇ 0.4%at its level · small run
τ³-bench lab68.4% ✎▁▁▂▁·▁▁·▁▁▂█ beats 16/25medium · avg 38%▇ 0.4%blunt · at its level · small run
coding▁▁▁▃▆██▇▆▄▃▁ #100 of 458 · 19% of its score · 7 tests

Beats Grok 4.3, Muse Glimmer 30B, Qwen3.5 Plus here, though they rank higher overall. Trails Llama 3.1 Nemotron Ultra 253B v1, Nemotron Cascade 2 30B-A3B, MiniMax-M2.1 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
SWE-bench Verified +468.6% ✎▄▄·····▄▄▄▄█ beats 6/7hard · avg 17%▇▇▇ 7.2%sharp · at its level · several runs
LiveCodeBench v6 (Feb–May 2025) lab +380.7% ✎▁▁▃▁█▅▄▇▅▅▃█ beats 32/38medium · avg 47%▇ 3.1%several runs
Arena WebDev (Code Arena)1357.3 ✎▁▂▄▃▅█▄▇▆▄▁▁ beats 31/119avg 981▇ 2.8%sharp
SWE-bench Multilingual +257.7% ✎▄·▄·····█▄█▄ beats 6/7hard · avg 18%▇ 2.5%at its level · several runs
SWE-bench Pro (split unspecified)42.1% ✎▄···▄···▄▄██ beats 6/7very hard · avg 6.5%▇ 1.6%sharp · at its level
SWE-bench Pro public set, Qwen-refined lab +151.2% ✎▄······▄█·▄▄ beats 3/5hard · avg 18%▇ 1.3%at its level · several runs
SciCode (level unspecified)25.6% ✎▄···▄▄█▄·▄·▄ beats 1/7very hard · avg 13%▇ 0.8%at its level
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #100 of 393 * · 1.9% of its score · 2 tests

Beats LongCat 2.0 (1.6T-A48B), Inkling Small (266B-A12B), GPT-5.4 nano here, though they rank higher overall. Trails Mistral Large 3 (675B-A41B), GPT-4.1, Gemma 4 12B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
LMArena (text)1408.7 ✎▁▁▁▂▃▃▃▆▆▇█▃ beats 253/379avg 1268▇ 1.5%blunt
IFEval (aggregation unspecified) lab95.0% ✎▁▁▃▁▂▄▄▆█▃▂▃ beats 139/141easy · avg 81%▇ 0.4%maxed out · small run
math▁▂▃▅▆▅▄█▄▁▁▁ #94 of 398 · 17% of its score · 12 tests

Beats Claude Sonnet 4.6, Claude Opus 4.5, MiniMax-M3 here, though they rank higher overall. Trails Gemini 2.5 Flash, GLM-4.5-Air, Granite 4.2 30B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HMMT February 2026 lab +184.3% ✎▃▃▂▃·██▆█▅▃▅ beats 25/33medium · avg 36%▇ 3.3%sharp · several runs
AIME 2026 lab +192.6% ✎▁▁▁▂▁▂▆█▆▂·▁ beats 32/49medium · avg 58%▇ 2.9%sharp · maxed out · small run · several runs
IMO-AnswerBench lab +179.9% ✎▁▁·▁▁▁▄▅█▄▂▃ beats 22/42medium · avg 36%▇ 2.2%small run · several runs
MathArena ArXivMath February 2026 (32 problems)31.3% ✎▁▁▁▁█▇▄▃▁▃·▃ beats 6/24very hard · avg 9.1%▇ 1.9%at its level
MathArena ArXivMath January 2026 (23 problems)53.3% ✎▂··▄▂▄█▆▆▄▄█ beats 7/23hard · avg 22%▇ 1.4%at its level
MATH-Vision lab +186.0% ✎▁▁·▁··▄▁▃▇▇█ beats 21/24medium · avg 54%▇ 1.3%small run · several runs
HMMT February 2025 lab92.0% ✎▁▁▁▃▁▅▅▃██▁▄ beats 21/31medium · avg 61%▇ 1.0%small run
MathArena ArXivMath December 2025 (17 problems)41.2% ✎▃·▃▃█▁·▄·▁▁▁ beats 8/18hard · avg 19%▇ 0.9%at its level
4 more tests
testresultvs the others in its rundifficultycountswhy
MathVista (testmini) lab87.8% ✎▁··▁▄▄▇█▅▆█▄ beats 46/47medium · avg 64%▇ 0.6%small run
HMMT November 2025 lab89.8% ✎▁▁····▁▅▃█▇▁ beats 13/23easy · avg 67%▇ 0.6%small run
PolyMATH lab71.2% ✎▁···▃▁·▇█▄▄▄ beats 22/23hard · avg 34%▇ 0.5%at its level · small run
DynaMath lab87.7% ✎▂·····▂··█▄▆ beats 10/10medium · avg 65%▇ 0.5%small run
vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #58 of 147 · 10% of its score · 21 tests

Beats DeepSeek V4 Flash Vision (Exp), GPT-5, MiniMax-M3 here, though they rank higher overall. Trails Gemma 4 26B-A4B, Qwen3.6 Flash, Qwen3.5 35B-A3B here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
MATH-Vision lab +186.0% ✎▁▁·▁··▄▁▃▇▇█ beats 21/24medium · avg 54%▇ 1.3%small run · several runs
Arena vision (style controlled)1216.4 ✎▁▁▂▂▂▂▄▅▅▆██ beats 83/145avg 1068▇ 1.1%blunt
OmniDocBench 1.5 overall (higher is better) lab +188.9% ✎▂··▂▂▂··▂▂█▅ beats 6/13easy · avg 70%▇ 1.0%small run · several runs
CharXiv (reasoning questions) lab79.5% ✎▂▅▃▅▅▆▆▅▆█▆▂ beats 75/100medium · avg 43%▇ 0.7%small run
MathVista (testmini) lab87.8% ✎▁··▁▄▄▇█▅▆█▄ beats 46/47medium · avg 64%▇ 0.6%small run
MMMU (val) lab82.3% ✎▁····▁▁▄█▇▆▆ beats 49/52medium · avg 58%▇ 0.6%small run
MMMU Pro lab75.0% ✎▂▃▁▃▆▇▃▇▇██▅ beats 38/52medium · avg 47%▇ 0.6%at its level · small run
BEAR-Bench (bilingual RU/EN enterprise and academic document images, overall accuracy)60.8% ✎▃▃·▅·▃·█·█▅▅ beats 8/14hard · avg 24%▇ 0.6%at its level
13 more tests
testresultvs the others in its rundifficultycountswhy
MMStar lab81.0% ✎▃▅▂▃▂▂▆█▆▆█▅ beats 30/34medium · avg 63%▇ 0.4%small run
HallusionBench lab +170.0% ✎▃▁·▃▃▃▄▅▄█▅▅ beats 31/32medium · avg 55%▇ 0.4%blunt · at its level · small run · several runs
PMC-VQA lab62.4% ✎▃···▃··▃▃·▃█ beats 5/7medium · avg 39%▇ 0.4%at its level · small run
CountBench lab97.8% ✎▂▂▃▂·▂▂▃▃·▅█ beats 18/18very easy · avg 92%▇ 0.4%maxed out · small run
SimpleVQA lab56.0% ✎▃·▃·▃·▃·█▅▃▃ beats 4/10hard · avg 34%▇ 0.4%at its level · small run
MMBench-EN dev v1.1 lab92.6% ✎█▄···▂··▄▂▄▆ beats 11/14easy · avg 82%▇ 0.3%blunt · maxed out · small run
ScreenSpot-Pro lab70.3% ✎▂▃▅▆██▅▂···▂ beats 18/24medium · avg 48%▇ 0.3%blunt · at its level · small run
RealWorldQA lab83.7% ✎▂▃▂█▃██▆█▂▆▆ beats 31/38easy · avg 72%▇ 0.2%blunt · small run
AI2D test lab +192.9% ✎▁▁▅▇▃▇·▄·█·▃ beats 23/28easy · avg 85%▇ 0.2%blunt · small run · several runs
SpaceCast-Bench predictive spatial reasoning (overall, 16 task types)39.3% ✎▃·▅▃▅▃··▅··█ beats 6/11very hard · avg 32%▇ 0.2%blunt · noisy
CC-OCR lab81.0% ✎▁·▂··▂▂▃▃▂▆█ beats 20/26easy · avg 72%▇ 0.1%blunt · small run
OCRBench lab89.4% ✎▁·▃▁▆▅▂█▅█▁▃ beats 29/37easy · avg 83%▇ 0.1%blunt · small run
PureDocBench document parsing (Avg3 over Clean/Digital/Real)69.6% ✎▂·▁·▂▂▅█·▁▂▂ beats 13/22medium · avg 62%▇ 0.0%blunt · at its level
web search▂▁▂▆▄█▃▃▁▁·▁ #62 of 80 · 3.5% of its score · 5 tests

Beats DeepSeek V4 Flash (Preview, Apr 2026), Grok 4.3, Nemotron 3 Ultra 550B-A55B here, though they rank higher overall. Trails GPT-4o mini (2024-07-18), Mistral Small 3.1 24B (2503), Llama 4 Scout (109B-A17B) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
WideSearch lab +161.1% ✎█·▂▆··▂···▂▂ beats 5/10hard · avg 33%▇ 0.9%at its level · small run · several runs
BrowseComp-ZH lab62.1% ✎▁▁▁▁▁▃·▁▄▅▄█ beats 16/23hard · avg 23%▇ 0.9%at its level · small run
BrowseComp lab61.0% ✎▂·▄▁▂▃▆▄█▂▃▄ beats 25/42hard · avg 23%▇ 0.7%at its level · small run
Humanity's Last Exam with tools lab48.5% ✎▂▂▂▁▄▂▃▆█▇▂▆ beats 29/54hard · avg 18%▇ 0.5%at its level · small run
SealQA Seal-0 lab47.2% ✎█····█·····█ beats 2/2hard · avg 19%▇ 0.4%at its level · small run
other tests46 not in a task · 40% of its score
testresultvs the others in its rundifficultycountswhy
GPQA Diamond lab +285.5% ✎▁▁▁▁▂▄▅▅███▄ beats 230/310hard · avg 45%▇▇ 6.2%sharp · several runs
Humanity's Last Exam lab +324.3% ✎▁▁▁▃▆▆▇██▆▁▃ beats 57/104very hard · avg 5.2%▇▇ 4.1%several runs
MMLU-Pro lab +386.1% ✎▁▁·▁▁▁▃▂▄▅█▁ beats 178/188medium · avg 56%▇ 3.6%several runs
NL2Repo lab +127.3% ✎▄·····█·▄▄·▄ beats 1/5very hard · avg 8.2%▇ 1.6%at its level · several runs
VitaBench lab +141.9% ✎▄··▂▄▆▆█▄▄▄▂ beats 20/21very hard · avg 11%▇ 1.3%at its level · small run · several runs
MedXPertQA MM lab62.4% ✎▂···▂·▃··▆▂█ beats 11/13very hard · avg 31%▇ 1.2%sharp · at its level · small run
QwenClawBench (Qwen internal) lab +152.2% ✎▄·▄····▄···█ beats 3/4hard · avg 22%▇ 1.2%at its level · several runs
KINA (Knowledge Index of Noah's Ark) accuracy, 899 ten-option items39.4% ✎▁▁▂▁▃▄█▆▄▅▂▃ beats 32/39very hard · avg 22%▇ 1.1%blunt · at its level
38 more tests
testresultvs the others in its rundifficultycountswhy
LongBench v2 lab60.6% ✎▂·▁·▂▂▁▃▄▁█▄ beats 22/28very hard · avg 35%▇ 0.9%at its level · small run
Claw-Eval Pass^3 lab +146.2% ✎█▄▄·▄·▄▄·▄·▄ beats 0/8hard · avg 24%▇ 0.9%at its level · several runs
LVBench lab +173.6% ✎▇·█▅▅▄▄··▃▁▁ beats 22/28medium · avg 52%▇ 0.8%at its level · small run · several runs
DeepPlanning lab +122.6% ✎▂·····▂··▄█▄ beats 4/9very hard · avg 8.7%▇ 0.8%small run · several runs
C-Eval lab90.5% ✎▁···▁▁▂▁▂▂▃█ beats 25/33medium · avg 70%▇ 0.8%small run
ZEROBench subquestions lab +136.2% ✎▄·▄···▄·▄██▄ beats 6/8hard · avg 16%▇ 0.8%at its level · small run · several runs
INCLUDE lab81.6% ✎▂····▂▂▄▄▄█▄ beats 32/36medium · avg 55%▇ 0.8%small run
SkillsBench 78-task subset (OpenCode, avg of 5) lab +127.2% ✎▄·····▄···▄█ beats 3/4very hard · avg 9.9%▇ 0.7%at its level · several runs
MMLU-Redux lab93.2% ✎▁▁▃▁▂▃▄▅▇▅█▄ beats 53/63easy · avg 78%▇ 0.7%small run
MMMLU lab85.9% ✎▂▁▁▂▃▂▄▄▅██▃ beats 37/46medium · avg 63%▇ 0.7%small run
OJBench lab40.1% ✎▁···▁▃·▅▄▅██ beats 24/26very hard · avg 13%▇ 0.7%at its level · small run
AndroidWorld lab +164.2% ✎▁·▁▁▂▃█▅▂▂·▂ beats 18/24medium · avg 41%▇ 0.6%blunt · at its level · small run · several runs
Video-MME (without subtitles) lab +182.8% ✎▁▃▁▃▁·█▄▄▅·▇ beats 25/27medium · avg 67%▇ 0.6%blunt · small run · several runs
Claw-Eval average score lab +164.3% ✎▄·▄··█▄▄▄▄▄▄ beats 4/9medium · avg 47%▇ 0.6%blunt · at its level · several runs
Global PIQA lab87.5% ✎▃▃··▃··▃▃▅█▅ beats 9/11medium · avg 72%▇ 0.6%small run
CPTUB average (SpeakLeash, 0-5 judge scale)86.7% ✎▁▂▂▂▃█▃▆▅▄▂▃ beats 92/95easy · avg 66%▇ 0.5%typical
IFBench (aggregation unspecified) lab76.5% ✎▃▇▂▄▅▅▂▄▄▆█▇ beats 68/80medium · avg 43%▇ 0.5%at its level · small run
MAXIFE lab88.0% ✎▂·▂·▂···▆·▄█ beats 10/11easy · avg 65%▇ 0.5%small run
MultiChallenge lab60.8% ✎▁·▁·▁▃▆▅▃▁█▅ beats 20/29hard · avg 30%▇ 0.5%at its level · small run
VideoMMMU lab82.3% ✎▂···▂···▂▂█▆ beats 7/10medium · avg 59%▇ 0.5%small run
SuperGPQA lab65.6% ✎▂▁▂▂▅▆▃▇▇▇█▃ beats 42/46medium · avg 36%▇ 0.5%at its level · small run
MMLU-ProX lab82.2% ✎▁·▁▁▁▂▁▃▂█▅▄ beats 27/35medium · avg 57%▇ 0.5%small run
VLMs Are Blind lab96.9% ✎▂·▂·····▃··█ beats 6/8very easy · avg 89%▇ 0.5%maxed out · small run
LingoQA lab82.0% ✎▄······▄·█·█ beats 5/5medium · avg 56%▇ 0.4%small run
SLAKE lab80.0% ✎▄·····▄▄·██▄ beats 6/7medium · avg 59%▇ 0.4%small run
Video-MME (with subtitles) lab87.0% ✎▁▁▁▁▁·▁▃▁▁▁█ beats 13/16medium · avg 71%▇ 0.4%small run
MMVU lab73.3% ✎▃···▃···▃▃▅█ beats 6/8medium · avg 48%▇ 0.4%at its level · small run
RefSpatial-Bench lab67.7% ✎▃·▃····▃▃▅▅█ beats 6/10medium · avg 40%▇ 0.4%at its level · small run
FullStackBench (Chinese) lab57.4% ✎█······█···█ beats 1/2hard · avg 29%▇ 0.4%at its level · small run
FullStackBench (English) lab60.1% ✎█····█·····█ beats 1/2hard · avg 31%▇ 0.4%at its level · small run
MLVU lab85.9% ✎▂···▂···▂▄▄█ beats 6/10easy · avg 69%▇ 0.4%small run
MMLongBench-Doc lab60.2% ✎▃····▃··▃▃▅█ beats 7/8medium · avg 35%▇ 0.4%at its level · small run
ERQA lab60.5% ✎▃··▃··█·▃█▅▅ beats 5/12medium · avg 36%▇ 0.4%at its level · small run
EmbSpatial-Bench lab84.5% ✎▁·····▁·▁·▃█ beats 9/10easy · avg 66%▇ 0.4%small run
NOVA-63 lab58.1% ✎▃▃·▃···▃█▃▃█ beats 9/11medium · avg 35%▇ 0.3%blunt · at its level · small run
MVBench lab74.6% ✎▁··▃▄▁▃▄▄▄█▄ beats 20/26medium · avg 66%▇ 0.1%blunt · at its level · small run
WMT24++ lab77.6% ✎▁·▁▁·▁·▁▂▆█▃ beats 13/27easy · avg 69%▇ 0.1%blunt · small run
Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate)87.9% ✎▁▂▁▂▅█▂▃▄▃▁▁ beats 23/98very easy · avg 90%▇ 0.0%blunt

$ where to run it

Through an API

api · Deep Infra · $0.84/M · $0.26 in / $2.6 out per 1M tokens · 262k context · deepinfra.com ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run Qwen3.5 27B · needs ~16.5 GB · from huggingface.co/lmstudio-community
    chat in the terminal
    llama cli -hf lmstudio-community/Qwen3.5-27B-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf lmstudio-community/Qwen3.5-27B-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run Qwen3.5 27B · needs ~16.5 GB · from huggingface.co/lmstudio-community

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/lmstudio-community/Qwen3.5-27B-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Qwen3.5 27B · needs ~17.18 GB · from ollama.com
    chat in the terminal
    ollama run qwen3.5:27b

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run Qwen3.5 27B · needs ~16.1 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/Qwen3.5-27B-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/Qwen3.5-27B-4bit

Official weights and documentation: huggingface.co/Qwen ↗

$ how the score was worked out · Qwen/Qwen3.5-122B-A10B model… 39%, Qwen/Qwen3.6-35B-A3B model ca… 17%, Qwen/Qwen3.6-27B model card 16% and 11 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

sourceresultsshare of the scorelicence
Qwen/Qwen3.5-122B-A10B model card · lab-reported71▇▇▇▇▇▇▇▇▇▇▇▇ 39%apache-2.0
Qwen/Qwen3.6-35B-A3B model card · lab-reported27▇▇▇▇▇ 17%apache-2.0
Qwen/Qwen3.6-27B model card · lab-reported16▇▇▇▇▇ 16%apache-2.0
Qwen/Qwen3.5-27B model card · lab-reported7▇▇ 8%apache-2.0
Kwaipilot/KAT-Coder-V2.5-Dev model card5▇▇ 7%Apache-2.0
LMArena leaderboard dataset3▇▇ 5%CC BY 4.0
MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets3▇ 4%CC BY-SA 4.0
ContractBench: Can LLM Agents Preserve Observation Contracts? (arXiv:2605.17281v2)1▇ 1%CC BY 4.0
Knowledge Index of Noah's Ark (arXiv:2606.05104v2)1▇ 1%CC BY 4.0
BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models (arXiv:2608.17895v1)1▇ 1%CC BY 4.0
SpeakLeash CPTUB (Complex Polish Text Understanding Benchmark)1▇ 1%MIT
SpaceCast-Bench: Evaluating Predictive Spatial Reasoning in Vision-Language Models (arXiv:2610.12402v1)1▇ 0%CC BY 4.0
How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean, Degraded, and Real-World Settings (arXiv:2605.07492v2)1▇ 0%CC BY 4.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)1▇ 0%Apache-2.0
prior · pull toward the middle▇ 0%
weighted average= 68.5

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit 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 · sources and licences · json