[whatmodel.sh](../../README.md) / Qwen3.5 9B
# Qwen3.5 9B
Alibaba · released 2026-03 · 9.65B params · apache-2.0
**score 62.0** ±0.4 · #181 of 732 · from 138 benchmark results across 25 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Qwen3.5 9B among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█●▅▅▄▄▂▁ #154 of 313 16% 15
coding ▁▁▁▃▅▇█●▅▄▃▁ #189 of 469 13% 9
chat & writing ▁▁▁▃▆▄▆█●▇▇▃ #150 of 402 1.0% 1
math ▁▁▃▅█▆▄●▄▂▁▁ #148 of 431 18% 15
vision & documents ▁▁▂▂▁▃▄▂●▇█▄ #73 of 153 8.5% 21
web search ▅·▅▅●·▃▃▃·▅▃ #9 of 15* 0.4% 1
agent tasks · #154 of 313 · 16% of its score
Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o4-mini](o4-mini.md), [Qwen3 235B-A22B Thinking 2507](qwen3-235b-a22b-thinking-2507.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Jamba 1.5 Large](jamba-1-5-large.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL V4 (December 2025 archive) 60.6% ▁▁▁▂▆▆▅▆█▂●▃ beats 100/114 medium · avg 42% 2.6% at its level · several runs
SWE-bench Pro (split unspecified) 38.0% ▂·▁▁▁▁▂●█▇▂▁ beats 18/58 very hard · avg 9.6% 2.1% sharp · at its level · small run · seve…
Terminal-Bench 2.1 21.7% ▁·▁·▁▁●▃▆█▅█ beats 7/63 very hard · avg 6.2% 2.0% sharp · several runs
SWE-bench Verified 50.0% ▁▁▁▁▁▁▂▄●▆▁▁ beats 57/138 hard · avg 23% 1.9% sharp · at its level · small run · seve…
ContractBench success rate (33 ob… 56.6% ▆··▆▄▂▄▂▆●█▄ beats 15/23 very hard · avg 13% 1.6% sharp · at its level
SWE-bench Multilingual 43.3% ▁··▁▃●▅██▂▂▁ beats 10/65 hard · avg 18% 1.0% at its level
OSWorld-Verified (maker) 41.8% █●·▂▅▂▃▂▅▃▆▆ beats 5/27 hard · avg 19% 0.8% at its level · small run
MCPMark 13.4% ▃·····●··▃█▃ beats 2/8 very hard · avg 5.5% 0.7% typical
Thinkingbox-bench pass@1 (507 sta… 5.7% ●··▃·▃·█▃▅██ beats 1/15 very hard · avg 5.1% 0.5% sharp
τ³-bench Banking 5.1% ▂▂·▅●▂▃▃▃▆▂▅ beats 5/24 very hard · avg 4.0% 0.5% blunt · too hard for it
BFCL v3 70.5% ▁▁▁▁▁▃▅█▆▅●▁ beats 153/171 medium · avg 55% 0.4% blunt · at its level · several runs
τ³-bench 9.3% ▃·▃·●·▃··▃·▃ beats 3/7 very hard · avg 13% 0.4% typical
τ²-bench aggregate (maker) 79.1% ▁▁▁·▁▁▃▃▃●█▁ beats 21/39 medium · avg 40% 0.4% at its level · noisy
BrowseComp 8.3% ▅···●·▃···▃▃ beats 3/7 very hard · avg 11% 0.4% noisy
Toolathlon-Verified 25.9% ●·▂▂▅▃···▂██ beats 0/18 hard · avg 24% 0.3% blunt · at its level · small run
coding · #189 of 469 · 13% of its score
Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [OpenThinker 32B](openthinker-32b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LiveCodeBench v6 (window unspecif… 63.7% ▂▃▄▄▅██●▄▇▄▆ beats 41/67 hard · avg 33% 3.7% at its level · several runs
SWE-bench Pro (split unspecified) 38.0% ▂·▁▁▁▁▂●█▇▂▁ beats 18/58 very hard · avg 9.6% 2.1% sharp · at its level · small run · seve…
SWE-bench Verified 50.0% ▁▁▁▁▁▁▂▄●▆▁▁ beats 57/138 hard · avg 23% 1.9% sharp · at its level · small run · seve…
Luau Leetcode fully solved (Torpe… 19.2% ▁▁●▁·▁▂▃▅▃█▇ beats 4/49 hard · avg 17% 1.8% sharp · at its level
LiveCodeBench v6 (Feb–May 2025) (… 65.6% ▁▁▃▁█▅●▇▅▅▃█ beats 16/38 medium · avg 41% 1.5% at its level · several runs
SWE-bench Multilingual 43.3% ▁··▁▃●▅██▂▂▁ beats 10/65 hard · avg 18% 1.0% at its level
HumanEval+ 89.6% ▁·▁▁▂▃▄▄█▅●▃ beats 130/145 easy · avg 80% 0.6% sharp · maxed out
MBPP+ 69.8% ▁··▁▁▃▄█●▄▁▁ beats 84/120 medium · avg 65% 0.4% typical
MultiPL-E 67.1% ▃··▄·▅▅●▄█▃▄ beats 16/31 medium · avg 54% 0.3% at its level
chat & writing · #150 of 402 · 1.0% of its score
Beats [INTELLECT-3](intellect-3.md), [MiniMax-M2](minimax-m2.md), [Nemotron 3 Super 120B-A12B](nemotron-3-super-120b-a12b.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md), [LFM2.5 2.6B](lfm2-5-2-6b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
IFEval (aggregation unspecified) 83.9% ▁▂▄▂▂▄▆██●▂▁ beats 127/145 easy · avg 80% 1.0% small run · several runs
math · #148 of 431 · 18% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md), [MiniMax-M2.7](minimax-m2-7.md) here, though they rank higher overall. Trails [Olmo 3 32B Think](olmo-3-32b-think.md), [Nemotron-Cascade-8B](nemotron-cascade-8b.md), [Olmo 3.1 32B Think](olmo-3-1-32b-think.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
HMMT February 2026 70.8% █·▄··▄··▄●·▄ beats 6/7 hard · avg 32% 2.8% sharp · at its level
MathArena ArXivMath February 2026… 27.3% ▁▁▁●█▇▄▃▁▃·▃ beats 3/24 very hard · avg 9.4% 1.8% at its level
MathArena ArXivMath January 2026 … 44.6% ▂··▄●▄█▆▆▄▄█ beats 3/23 hard · avg 23% 1.7% at its level
OTIS mock AIME 2024–2025 42.8% ▃▃▄▃▃●▇▇█▃▂▅ beats 72/197 very hard · avg 7.2% 1.6% sharp · at its level · noisy
IMO-AnswerBench 69.8% ▄▄···▄·▄··●▄ beats 5/6 hard · avg 32% 1.5% at its level
AIME 2026 88.2% ▁▁▁▂▂▂●█▆▂·▁ beats 17/52 medium · avg 53% 1.3% sharp
AIME 2025 56.1% ▁·▁▁▁▂▅●█▇▅▁ beats 42/133 hard · avg 30% 1.3% sharp · at its level
MATH-Vision (maker) 78.9% ▁▁·▁··▄▁▃▇●█ beats 14/24 medium · avg 54% 1.2% at its level · small run · several runs
MathArena ArXivMath December 2025… 39.7% ▃·▃▃●▁·▄·▁▁▁ beats 6/18 hard · avg 19% 1.1% at its level
HMMT February 2025 (maker) 83.2% ▁▁▁▃▁▅▅●██▁▄ beats 14/31 medium · avg 49% 0.9% typical
HMMT November 2025 (maker) 82.9% ▁▁····▁▅●█▇▁ beats 7/23 medium · avg 62% 0.8% typical
PolyMATH (maker) 57.3% ▁···▃▁·▇█●▄▄ beats 16/23 hard · avg 34% 0.6% at its level · small run
MathVista (testmini) (maker) 85.7% ▁··▁▄▄▆▇▄▅●▄ beats 40/48 medium · avg 68% 0.6% small run
DynaMath (maker) 83.6% ▂·····▂··●▄▆ beats 5/10 easy · avg 69% 0.4% small run
We-Math (maker) 75.2% ●··········▃ beats 1/3 medium · avg 58% 0.4% small run
vision & documents · #73 of 153 · 8.5% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5](gpt-5.md), [Inkling Small (266B-A12B)](inkling-small.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
MATH-Vision (maker) 78.9% ▁▁·▁··▄▁▃▇●█ beats 14/24 medium · avg 54% 1.2% at its level · small run · several runs
ChartArena chart parsing (English… 40.6% ▁·▁·▃▃▁▇●█·▄ beats 13/22 hard · avg 27% 1.1% blunt · at its level
BEAR-Bench (bilingual RU/EN enter… 49.6% ▃▃·▅·▃·●·█▅▅ beats 7/14 hard · avg 24% 0.7% at its level
CharXiv (reasoning questions) (ma… 73.0% ▂▅▃▅▅▆▆●▆█▆▂ beats 63/101 medium · avg 46% 0.7% at its level · small run
MathVista (testmini) (maker) 85.7% ▁··▁▄▄▆▇▄▅●▄ beats 40/48 medium · avg 68% 0.6% small run
MMMU (val) (maker) 78.4% ▁····▁▁▃█▆▆● beats 45/53 medium · avg 60% 0.6% at its level · small run
MMMU Pro (maker) 70.1% ▂▃▁▃▆▇▃▇●██▅ beats 34/52 medium · avg 48% 0.5% at its level · small run
OmniDocBench 1.5 overall (higher … 87.7% ▂··▂▂▂··▂●█▅ beats 5/13 easy · avg 73% 0.4% small run
MMStar (maker) 79.7% ▃▅▂▃▂▂▆██▆●▅ beats 29/35 medium · avg 65% 0.4% at its level · small run
CountBench (maker) 97.2% ▂▂▃▂·▂▂▃▃·▅● beats 14/18 very easy · avg 92% 0.4% maxed out · small run
PMC-VQA (maker) 57.9% ▃···▃··▃●·▃█ beats 3/7 medium · avg 42% 0.3% at its level · small run
SimpleVQA (maker) 51.2% ▃·▃·▃·●·█▅▃▃ beats 3/10 hard · avg 34% 0.3% blunt · at its level · small run
MMBench-EN dev v1.1 (maker) 90.1% █▄···▂··●▂▄▆ beats 8/14 easy · avg 82% 0.3% blunt · small run
ScreenSpot-Pro (maker) 65.2% ▂▃▅▆█●▅▂···▂ beats 15/24 medium · avg 48% 0.3% blunt · at its level · small run
HallusionBench (maker) 69.3% ▂▁·▂▂▂▃▅▃█▅● beats 30/33 medium · avg 56% 0.2% blunt · at its level · small run
SpaceCast-Bench predictive spatia… 34.6% ▃·●▃▅▃··▅··█ beats 2/11 very hard · avg 33% 0.2% noisy
RealWorldQA (maker) 80.3% ▁▃▁▇▃▇▇▅●▁▅▅ beats 30/39 easy · avg 72% 0.2% blunt · small run
CC-OCR (maker) 79.3% ▁·▂··▂▂▃▃●▆█ beats 13/26 easy · avg 72% 0.1% blunt · small run
OCRBench (maker) 89.2% ▁·▃▁▆▅▂█▅●▁▃ beats 27/37 easy · avg 83% 0.1% blunt · small run
AI2D test (maker) 90.2% ▁▁▅▇▃▇·●·█·▃ beats 20/28 easy · avg 85% 0.1% blunt · small run
PureDocBench document parsing (Av… 70.9% ▂·▁·▂▂▅●·▁▂▂ beats 17/22 medium · avg 62% 0.0% blunt · at its level
web search · #9 of 15 · 0.4% of its score
Trails [Spark-X2.5 1.7B](spark-x2-5-1-7b.md), [Granite 4.2 3B](granite-4-2-3b.md), [Spark-X2.6 1.7B](spark-x2-6-1-7b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BrowseComp 8.3% ▅···●·▃···▃▃ beats 3/7 very hard · avg 11% 0.4% noisy
other tests (53)
test result vs the others in its run difficulty counts why
GPQA Diamond 79.0% ▁▁▁▁▂▄▆▄▅●█▅ beats 124/211 hard · avg 39% 14% sharp · at its level · several runs
MMLU-Redux 91.1% ▁▁▃▁▂▃▄▅▇●█▄ beats 43/63 easy · avg 78% 2.8% several runs
Humanity's Last Exam 14.3% ▄▄·▄··▄·▄·▄● beats 7/7 very hard · avg 3.8% 2.4% typical
MMLU-Pro (maker) 82.5% ▁▁·▁▁▁▃▂▄▅●▁ beats 147/188 medium · avg 61% 1.9% several runs
GPQA (subset unspecified) 77.2% █···▃···▅·▃● beats 7/7 hard · avg 44% 1.7% sharp · at its level
IFBench (aggregation unspecified) 64.5% ▂▆▂▄▄▄▂▄●▆█▇ beats 47/86 medium · avg 39% 1.7% at its level · several runs
Polish medical board exams, bias-… 63.1% ▄·▂▂▂··█▂●·▄ beats 10/13 hard · avg 46% 1.4% at its level
KINA (Knowledge Index of Noah's A… 30.1% ▁▁▂▁▃▄█●▄▅▂▃ beats 22/39 very hard · avg 21% 1.3% typical
LongBench v2 (maker) 55.2% ▂·▁·▂▃▁▄▄▃●▄ beats 20/33 very hard · avg 35% 1.2% at its level
C-Eval (maker) 88.2% ▁···▁▁▂▁▂▂●█ beats 19/33 medium · avg 73% 1.1% typical
CPTUB average (SpeakLeash, 0-5 ju… 65.5% ▁▂▂▂▃█●▆▅▄▂▃ beats 45/95 medium · avg 65% 1.0% typical
MMMLU (maker) 81.2% ▂▁▁▂▃▂▄▄▅●█▃ beats 30/46 medium · avg 63% 1.0% at its level
MedXPertQA MM (maker) 49.9% ▂···▂·▃··●▂█ beats 7/13 very hard · avg 30% 1.0% sharp · at its level · small run
KOPA-Bench (pass@1) 32.8% ▄·█··▄··●▄·█ beats 4/8 very hard · avg 14% 1.0% at its level
RobloxQA v2.0 multiple choice (To… 74.5% ●▁▃▁█▄▃▃▆▂▂▂ beats 0/49 easy · avg 77% 1.0% typical
INCLUDE (maker) 75.6% ▂····▂▂▄▄▄●▄ beats 22/36 medium · avg 56% 0.9% at its level · small run
EuroEval Life in the UK (English … 89.8% ▂·▂▂·▂·▃●·█▃ beats 7/14 easy · avg 80% 0.7% typical
GSM-Plus 87.9% ▁··▁▁▁·▃▁▄█● beats 16/16 easy · avg 84% 0.7% blunt · several runs
Global PIQA (maker) 83.2% ▃▃··▃··▃▃●█▅ beats 5/11 medium · avg 71% 0.7% at its level
MultiChallenge (maker) 54.5% ▁·▁·▁▃▆▅●▂█▅ beats 17/30 hard · avg 31% 0.7% at its level
VitaBench (maker) 29.8% ▄··▂▄▆▆●▄▄▄▂ beats 14/21 very hard · avg 11% 0.7% typical
MAXIFE (maker) 83.4% ▂·▂·▂···▆·●█ beats 6/11 medium · avg 64% 0.7% typical
MMLU-ProX (maker) 76.3% ▁·▁▁▁▂▁▃●█▅▄ beats 13/36 medium · avg 56% 0.6% at its level
EuroEval HellaSwag (English), val… 82.1% ▂··▂▂·▂▆▆●·▄ beats 12/15 medium · avg 63% 0.6% at its level
OJBench (maker) 29.2% ▁···▁▃·▅▄▅●█ beats 15/26 very hard · avg 13% 0.6% at its level · small run
Next Steps data preparation (bala… 62.1% ▄·●···█·▄▄·▄ beats 1/6 medium · avg 52% 0.6% at its level
BFCL multi-turn (version not stat… 52.3% ▄▄···▄·▄●▄▄▄ beats 4/8 hard · avg 26% 0.5% at its level
Chess Puzzles (Epoch) 2.0% ●▁▁▁▁▂▃▃▃▂▁▁ beats 22/137 very hard · avg 5.8% 0.5% sharp · too hard for it · noisy
NOVA-63 (maker) 55.9% ▃▃·▃···▃█●▃█ beats 7/11 medium · avg 39% 0.5% blunt · at its level
SuperGPQA (maker) 58.2% ▂▁▂▂▅▆▃▇▇●█▃ beats 31/46 medium · avg 37% 0.4% at its level · small run
VideoMMMU (maker) 78.9% ▂···▂···▂●█▆ beats 3/10 medium · avg 62% 0.4% at its level · small run
VLMs Are Blind (maker) 93.7% ▂·▂·····●··█ beats 3/8 very easy · avg 87% 0.4% maxed out · small run
LVBench (maker) 70.0% ▇·█▅▅●▄··▃▁▁ beats 19/28 medium · avg 54% 0.4% at its level · small run
Claw-Eval average score 66.5% ▄·▄··█▄▄●▄▄▄ beats 6/9 medium · avg 53% 0.4% blunt · at its level
Video-MME (with subtitles) (maker) 84.5% ▁▁▁▁▁·▁▃▁▁●█ beats 10/16 medium · avg 73% 0.4% small run
SLAKE (maker) 79.0% ▄·····▄▄·●█▄ beats 4/7 medium · avg 64% 0.4% small run
ERQA (maker) 55.5% ▃··▃··●·▃█▅▅ beats 4/12 medium · avg 38% 0.4% at its level · small run
MMVU (maker) 67.8% ▃···▃···▃●▅█ beats 3/8 medium · avg 50% 0.4% at its level · small run
RefSpatial-Bench (maker) 58.5% ▃·▃····▃●▅▅█ beats 3/10 medium · avg 40% 0.4% at its level · small run
LingoQA (maker) 80.4% ▄······▄·●·█ beats 2/5 medium · avg 61% 0.4% small run
MLVU (maker) 84.4% ▂···▂···▂●▄█ beats 4/10 easy · avg 72% 0.3% small run
ZEROBench subquestions (maker) 31.1% ▄·▄···▄·●██▄ beats 3/8 hard · avg 17% 0.3% at its level · small run
MMLongBench-Doc (maker) 57.7% ▃···▃···▃●█▅ beats 4/9 medium · avg 39% 0.3% at its level · small run
AndroidWorld (maker) 57.8% ▁·▁▁▂▃●▅▂▂·▂ beats 13/24 medium · avg 41% 0.3% at its level · small run
NTCIR SciClaimEval Subtask 1 pair… 48.3% ●▃···▃·▃▃█·▃ beats 0/8 medium · avg 45% 0.3% blunt · at its level
EmbSpatial-Bench (maker) 83.0% ▁·····▁·▁·●█ beats 3/10 easy · avg 70% 0.3% small run
AutomationBench v1.0.6 5.0% ●···▃·▃▃··█▅ beats 0/8 very hard · avg 6.9% 0.3% noisy · small run
Video-MME (without subtitles) (ma… 78.4% ▁▃▁▃▁·█▄▄●·▇ beats 20/27 medium · avg 67% 0.3% blunt · at its level · small run
DeepPlanning (maker) 18.0% ▂·····▂··●█▄ beats 3/9 very hard · avg 7.1% 0.2% noisy
MVBench (maker) 74.4% ▁··▃▄▁▃▄▄▄●▄ beats 18/26 medium · avg 66% 0.1% blunt · at its level · small run
Multi-IF (aggregation unspecified) 62.5% ▁▁▁▁▁●▃▄█▄▁▂ beats 7/41 medium · avg 63% 0.1% noisy
LIBRA Mini (Russian long context,… 41.5% ▄···▄█·●█··▄ beats 4/8 medium · avg 38% 0.1% blunt · at its level
WMT24++ (maker) 72.6% ▁·▁▁·▁·▁●▆█▃ beats 6/27 easy · avg 66% 0.1% blunt · at its level
source results share of the score
Qwen/Qwen3.5-9B model card (submitted /… 62 31% ▇▇▇▇
XHToken/Spark-X2.5-4B model card 10 13% ▇▇
JetBrains/Mellum2.1-12B-A2.5B-Thinking … 11 9% ▇
JetBrains/Mellum2-12B-A2.5B-Instruct mo… 8 7% ▇
Epoch AI 3 5% ▇
Qwen/Qwen3.5-4B model card (submitted /… 4 5% ▇
MathArena ArXivMath / BrokenArXiv / ArX… 3 5% ▇
LiquidAI/LFM2.5-2.6B model card 7 4% ▇
XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B mo… 5 4% ▇
JetBrains/Mellum2-12B-A2.5B-Thinking mo… 7 3% ▇
Roblox LLM Leaderboard 2 3% ▇
ContractBench: Can LLM Agents Preserve … 1 2% ▇
Multi-Step Tool-Calling over Korean Ope… 2 1% ▇
Reassessing High-Performing LLMs on Pol… 1 1% ▇
EuroEval leaderboards 2 1% ▇
Knowledge Index of Noah's Ark 1 1% ▇
ChartArena: Benchmarking Chart Parsing … 1 1% ▇
SpeakLeash CPTUB 1 1% ▇
BEAR-Bench: A Bilingual Enterprise and … 1 1% ▇
Agentic coding without the cloud: evalu… 1 1% ▇
One Success Isn't Reliability: Thinking… 1 1% ▇
SciTrue: Reliable Scientific Claim Vali… 1 0% ▇
SpaceCast-Bench: Evaluating Predictive … 1 0% ▇
ai-forever LIBRA leaderboard 1 0% ▇
How Far Is Document Parsing from Solved… 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 62.0
Licences: Qwen/Qwen3.5-9B model card ([apache-2.0](../../licenses/Qwen--Qwen3.5-9B.txt)) · XHToken/Spark-X2.5-4B model card ([Apache-2.0](https://huggingface.co/XHToken/Spark-X2.5-4B/blob/0bcb35678590218655dff3765b9e61c83b35e9c4/README.md)) · JetBrains/Mellum2.1-12B-A2.5B-Thinking model card ([Apache-2.0](https://huggingface.co/JetBrains/Mellum2.1-12B-A2.5B-Thinking/blob/92ddae9fc7665e9f801d141d2e5a6b2caf2460c4/README.md)) · JetBrains/Mellum2-12B-A2.5B-Instruct model card ([Apache-2.0](https://huggingface.co/JetBrains/Mellum2-12B-A2.5B-Instruct/blob/273d9c1c6882ac8e1562e5b7e043857d35a787b6/README.md)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Qwen/Qwen3.5-4B model card ([apache-2.0](../../licenses/Qwen--Qwen3.5-4B.txt)) · MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets ([CC BY-SA 4.0](../../licenses/matharena-sharealike-NOTICE.txt)) · LiquidAI/LFM2.5-2.6B model card ([LFM Open License v1.0](https://huggingface.co/LiquidAI/LFM2.5-2.6B/blob/654f9463ce32b05d0429d76fe1f580b27d4c1ac0/LICENSE)) · XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B model card ([MIT](https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B/blob/2367e865d009c13ac81713a2878291d33ab28177/README.md)) · JetBrains/Mellum2-12B-A2.5B-Thinking model card ([Apache-2.0](https://huggingface.co/JetBrains/Mellum2-12B-A2.5B-Thinking/blob/a7311550557e93cc706ab5dd3d879c1a11703ab4/README.md)) · Roblox LLM Leaderboard ([MIT](../../licenses/roblox-llm-leaderboard-MIT.txt)) · ContractBench: Can LLM Agents Preserve Observation Contracts? ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Multi-Step Tool-Calling over Korean Open Public APIs: A Benchmark and a Data-Synthesis Recipe ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance? ([CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · Knowledge Index of Noah's Ark ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · ChartArena: Benchmarking Chart Parsing across Languages ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · SpeakLeash CPTUB ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Agentic coding without the cloud: evaluating open-weight LLMs on longitudinal data preparation tasks ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · One Success Isn't Reliability: Thinkingbox ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · SciTrue: Reliable Scientific Claim Validation with Frontier and Open Language Models at the NTCIR SciClaimEval Task ([CC BY 4.0](../../licenses/paper-2609.00654v1-CC-BY-4.0.txt)) · SpaceCast-Bench: Evaluating Predictive Spatial Reasoning in Vision-Language Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · ai-forever LIBRA leaderboard ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### Through an API
api (Deep Infra) · $0.11/M blended · $0.1 in / $0.15 out per 1M tokens · [deepinfra.com](https://deepinfra.com/models)
[price source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/providers/deepinfra/models/Qwen/Qwen3.5-9B.toml): Standard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/alibaba/qwen3.5-9b.toml)
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Qwen3.5 9B (Q4_K_M, needs ~5.6 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3.5-9B-GGUF/tree/1379f25c6b505a3fc737bd7818cb09389cf807c1))
```sh
llama cli -hf lmstudio-community/Qwen3.5-9B-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf lmstudio-community/Qwen3.5-9B-GGUF:Q4_K_M
```
Other versions: Q8_0 (~9.5 GB) `llama cli -hf lmstudio-community/Qwen3.5-9B-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Qwen3.5 9B (Q4_K_M, needs ~5.6 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3.5-9B-GGUF/tree/1379f25c6b505a3fc737bd7818cb09389cf807c1))
```sh
lms get https://huggingface.co/lmstudio-community/Qwen3.5-9B-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~9.5 GB) `lms get https://huggingface.co/lmstudio-community/Qwen3.5-9B-GGUF@Q8_0 -y` · 4bit (~6 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3.5-9B-MLX-4bit -y` · 8bit (~10.4 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Qwen3.5-9B-MLX-8bit -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run Qwen3.5 9B (Q4_K_M, needs ~6.29 GB, from [ollama.com](https://ollama.com/library/qwen3.5/tags))
```sh
ollama run qwen3.5:9b
```
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Qwen3.5 9B (4bit, needs ~6 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Qwen3.5-9B-MLX-4bit/tree/b455506b0f574c74616dbcd56879bde38fafcff3))
```sh
mlx_lm.chat --model lmstudio-community/Qwen3.5-9B-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model lmstudio-community/Qwen3.5-9B-MLX-4bit
```
Other versions: 8bit (~10.4 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Qwen3.5-9B-MLX-8bit`
Model repository and documentation: [huggingface.co/Qwen](https://huggingface.co/Qwen/Qwen3.5-9B)
## Published speed measurements
Reference results for the hardware, engine and settings below. These are not verified speeds of the downloads above or predictions for your machine.
hardware precision speed engine and settings
Ryzen 7 H 255 (Radeon 780M)… Q4_K_M 14.86 tok/s llama.cpp · whatmodel bench.py 3.…
Apple M3 Pro, 36 GB q4_k_m 21.93 tok/s ollama (version not recorded) · l…
RTX 2080 q4_k_m 44.68 tok/s ollama (version not recorded) · l…
Each speed links to its source, which has the full settings.
## Quantization evidence
How much of the full-precision result a quantized version kept, measured within one study. Rows from the same study share a table and aren't independent. They don't describe every download of that precision.
precision test full → quantized kept
FP8 W8A8 dynamic (E4M3; weigh… MMLU-Pro 0-shot (lm-evaluation-ha… 0.825 → 0.824 99.9%
FP8 W8A8 dynamic (E4M3; weigh… GSM8K-Platinum 0-shot (lm-evaluat… 0.947 → 0.945 99.8%
FP8 W8A8 dynamic (E4M3; weigh… GPQA Diamond 0-shot (lighteval, v… 0.84 → 0.838 99.8%
FP8 W8A8 dynamic (E4M3; weigh… MATH-500 0-shot (lighteval, vLLM;… 0.85 → 0.847 99.6%
FP8 W8A8 dynamic (E4M3; weigh… AIME 2025 0-shot (lighteval, vLLM… 0.883 → 0.879 99.5%
FP8 W8A8 dynamic (E4M3; weigh… IFEval prompt-level strict 0-shot… 0.903 → 0.889 98.4%
INT4 W4A16 (GPTQ, symmetric, … GSM8K-Platinum 0-shot (lm-evaluat… 0.944 → 0.946 100.2%
INT4 W4A16 (GPTQ, symmetric, … MMLU-Pro 0-shot (lm-evaluation-ha… 0.824 → 0.817 99.2%
INT4 W4A16 (GPTQ, symmetric, … MATH-500 0-shot (lighteval, vLLM;… 0.852 → 0.838 98.4%
INT4 W4A16 (GPTQ, symmetric, … IFEval prompt-level strict 0-shot… 0.895 → 0.877 98.0%
INT4 W4A16 (GPTQ, symmetric, … GPQA Diamond 0-shot (lighteval, v… 0.822 → 0.779 94.8%
INT4 W4A16 (GPTQ, symmetric, … AIME 2025 0-shot (lighteval, vLLM… 0.854 → 0.688 80.6%
INT8 W8A8 (weights static per… GPQA Diamond 0-shot (lighteval, v… 0.822 → 0.823 100.1%
INT8 W8A8 (weights static per… MATH-500 0-shot (lighteval, vLLM;… 0.852 → 0.853 100.1%
INT8 W8A8 (weights static per… IFEval prompt-level strict 0-shot… 0.895 → 0.896 100.1%
INT8 W8A8 (weights static per… AIME 2025 0-shot (lighteval, vLLM… 0.854 → 0.854 100.0%
INT8 W8A8 (weights static per… GSM8K-Platinum 0-shot (lm-evaluat… 0.944 → 0.943 99.9%
INT8 W8A8 (weights static per… MMLU-Pro 0-shot (lm-evaluation-ha… 0.824 → 0.82 99.5%
Each kept value links to its source.
[interactive version](https://whatmodel.sh/model/qwen3-5-9b) · [JSON](https://whatmodel.sh/api?format=json&all=1)