The LLM benchmark aggregator.

Qwen3 235B A22B Instruct

Alibaba · 235B · A22B · open weights

Strongest in Safe-prompt compliance (#5 of 82), weakest in Jailbreak resistance (#213 of 268). Above par in 21 of 27 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5 and Claude Fable 5.1 and ahead of O4 Mini and Grok 4.1 Fast.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Coding54.5−11.7#42/1792/6

Code generation56.3−9#10/201/1

Agentic coding53.7−13.2#50/1731/5

Agents54.3−12.1#64/2981/6

Tool use60.4−13.6#19/811/1

Safety54−7.4#72/3293/3

Safe-prompt compliance59.9−0.5#5/821/1

Secure code62.1−4.1#45/2701/1

Factual grounding51.6−19.1#46/1041/1

Harm refusal54.5−7#78/2922/2

Fairness53.9−16.1#78/2922/2

Toxicity avoidance53−6#150/2681/1

Jailbreak resistance40.4−24.6#213/2681/1

Professional51.8−10.6#76/1941/2

Medical53.8−12.7#57/1491/1

Legal51.4−14.5#80/1731/1

Finance51.2−11.2#88/1621/1

Core abilities50.8−14.5#89/1991/4

General intelligence51.1−18.3#77/1781/3

Human preference58.8−10#108/3541/1

Knowledge50−12.5#112/1971/3

Academic knowledge50−13.9#78/1231/1

Reasoning45.9−19.5#130/1953/5

Mathematics54.7−9.1#58/1521/2

Science45.9−17.6#88/1221/1

Reasoning38.9−30.1#124/1472/3

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for Qwen3 235B A22B Instruct, left for the other.

Gemini 2.0 Flashwon 22 · lost 1

Mistral Large 3won 18 · lost 1

Gemini 1.5 Flashwon 16 · lost 1

DeepSeek V3won 24 · lost 2

GPT-4o miniwon 22 · lost 2

Gemini 1.5 Prowon 17 · lost 2

Command R+won 16 · lost 2

GPT-4.1 nanowon 22 · lost 3

GPT-4.1 miniwon 22 · lost 3

Command Awon 20 · lost 3

Llama 3.3 70B Instructwon 17 · lost 3

Llama 4 Scout Instructwon 20 · lost 4

Llama 4 Maverick Instructwon 21 · lost 5

Gemini 2.5 Flash-Litewon 16 · lost 4

GPT OSS 20Bwon 16 · lost 5

Gemma 4 31Bwon 15 · lost 5

GPT-4owon 18 · lost 6

Claude 3.5 Haikuwon 14 · lost 5

Kimi K2won 19 · lost 7

Ling 3.0 Flashwon 12 · lost 5

Inkling Smallwon 12 · lost 5

GPT-5.4 nanowon 13 · lost 6

Claude 3.5 Sonnetwon 14 · lost 7

Grok 4.3won 14 · lost 7

Grok 3won 16 · lost 8

NVIDIA Nemotron 3 Ultra 550B A55Bwon 15 · lost 8

GPT OSS 120Bwon 16 · lost 9

Kimi K2.6won 14 · lost 8

Claude Haiku 4.5won 15 · lost 9

GPT-4.1won 16 · lost 10

MiMo V2.5 Prowon 12 · lost 8

Qwen3.7 Maxwon 13 · lost 9

Gemini 3.5 Flash-Litewon 10 · lost 7

GPT-5 nanowon 14 · lost 10

GLM-4.5won 11 · lost 8

MiniMax M2.7won 12 · lost 9

GPT-6 Lunawon 12 · lost 9

O3 Miniwon 13 · lost 10

DeepSeek V4 Flashwon 13 · lost 10

Grok 4won 14 · lost 11

DeepSeek V3.2won 14 · lost 11

MiniMax M2.5won 10 · lost 8

Grok 4 Fastwon 10 · lost 8

Gemini 3.1 Flash-Litewon 10 · lost 8

GLM-4.6won 13 · lost 11

GLM-4.7won 9 · lost 8

GLM-5.1won 11 · lost 10

Grok 4.1 Fastwon 12 · lost 11

O4 Miniwon 14 · lost 13

GPT-5.4 miniwon 9 · lost 9

Gemini 3.6 Flashwon 11 · lost 11

Claude Sonnet 4won 13 · lost 13

Gemini 2.5 Flashwon 13 · lost 14

DeepSeek R1won 12 · lost 13

MiniMax M3won 11 · lost 12

GLM-5.3 Flashwon 11 · lost 12

MiMo V2.6 Flashwon 9 · lost 10

Qwen3.8 27Bwon 8 · lost 9

Claude 3.7 Sonnetwon 10 · lost 12

Mistral Large 4won 9 · lost 11

GLM-5.2won 10 · lost 13

Qwen3.8 Maxwon 10 · lost 13

Kimi K2.5won 10 · lost 14

Gemini 3.5 Flashwon 9 · lost 13

Inklingwon 9 · lost 14

GPT-5.6 Lunawon 9 · lost 14

Grok 3 Miniwon 8 · lost 13

Claude Sonnet 4.6won 8 · lost 13

DeepSeek V4 Prowon 9 · lost 15

Kimi K2 Thinkingwon 7 · lost 13

MiMo V2.6 Prowon 7 · lost 13

Gemini 3.8 Flashwon 8 · lost 15

GPT-5 miniwon 9 · lost 17

GLM-5won 6 · lost 12

Gemini 3.1 Prowon 6 · lost 12

O1won 7 · lost 14

Gemini 3 Prowon 7 · lost 16

GPT-6 Solwon 6 · lost 15

Claude Opus 4.6won 6 · lost 15

Gemini 2.5 Prowon 7 · lost 19

GPT-5.1won 6 · lost 17

Grok 4.5won 6 · lost 17

Claude Opus 4won 6 · lost 19

Kimi K3won 4 · lost 13

GPT-5.6 Terrawon 5 · lost 18

GPT-5.4won 5 · lost 19

Claude Opus 4.5won 4 · lost 16

GPT-5won 5 · lost 20

Claude Sonnet 4.5won 5 · lost 21

Muse Spark 1.1won 4 · lost 17

GPT-5.6 Solwon 4 · lost 20

Claude Opus 4.7won 4 · lost 20

Claude Sonnet 5won 3 · lost 19

Grok 4.6won 3 · lost 20

GPT-5.2won 3 · lost 22

GPT-6 Astrawon 2 · lost 15

DeepSeek V4.1 Flashwon 2 · lost 15

O3won 3 · lost 23

GPT-5.5won 2 · lost 21

GLM-5.3won 1 · lost 16

Grok 4.7won 1 · lost 19

Claude Opus 5won 1 · lost 21

Claude Opus 5.5won 1 · lost 21

Claude Opus 4.8won 1 · lost 22

Claude Fable 5.1won 0 · lost 17

Claude Fable 5won 0 · lost 17

§ 3 · Sources

Where the numbers come from

11 publications, 25 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1422

ARC-AGI-2 ↗1 measureread 2026-11-10

ARC-AGI-2 1.3%

LiveCodeBench ↗1 measureread 2026-11-10

LiveCodeBench 65.9%

Aider polyglot ↗1 measureread 2026-11-10

Aider polyglot 59.6%

BFCL ↗1 measureread 2026-11-10

BFCL v4 52.15%

Kagi LLM Benchmark ↗1 measureread 2026-11-10

Kagi LLM Benchmark 55%

SimpleBench ↗1 measureread 2026-11-10

SimpleBench 31%

Vals.ai ↗6 measuresread 2026-11-10

Vals · LegalBench 80.18%Vals · TaxEval 70.65%Vals · MedQA 90.62%Vals · GPQA Diamond 70.2%Vals · MMLU Pro 81.25%Vals · AIME 83.96%

HELM Safety ↗5 measuresread 2026-11-10

HELM Safety · HarmBench 79.6%HELM Safety · SimpleSafetyTests 100%HELM Safety · Anthropic Red Team 99.9%HELM Safety · BBQ 96.2%HELM Safety · XSTest 98.6%

Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-11-10

Enkrypt · Jailbreak risk 20.9%Enkrypt · Harmful content risk 20%Enkrypt · CBRN risk 15.3%Enkrypt · Toxicity risk 4.1%Enkrypt · Bias risk 77.8%Enkrypt · Insecure code risk 8.4%

Vectara Hallucination Leaderboard ↗1 measureread 2026-11-10

Vectara · Factual consistency 90.7%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=qwen3-235b-a22b-instruct)](https://publicai.io/model-index/m/qwen3-235b-a22b-instruct)