Request a pilot
‹ PublicAI Index

The LLM benchmark aggregator.

Mixtral 8x22b Instruct

Mistral · 141B · open weights

Strongest in Safe-prompt compliance (#57 of 82), weakest in Human preference (#268 of 346). Above par in 2 of 11 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4 and Claude Sonnet 4 and ahead of Llama 3.1 70B Instruct and Gemma 2 9B It.

§ 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.

Professional45.8#139/187
Medical38.6#135/149
Safety48#234/331
Safe-prompt compliance49.9#57/82
Toxicity avoidance55.2#100/271
Jailbreak resistance55.7#106/271
Fairness47.4#149/296
Secure code46.3#188/273
Harm refusal45.2#241/296
Human preference40.5#268/346
Human preference40.5#268/346

§ 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 Mixtral 8x22b Instruct, left for the other.

OLMoE 1B-7B Instructwon 7 · lost 0
Smollm2 1.7B Instructwon 8 · lost 0
Zephyr 7B Betawon 8 · lost 0
Nous Hermes 2 Mixtral 8x7b Dpowon 8 · lost 0
Mixtral 8x7b Instructwon 11 · lost 0
DBRX Instructwon 8 · lost 1
Jamba 1.6 Miniwon 7 · lost 1
Jamba 1.6 Largewon 7 · lost 1
Gemma 2B Itwon 7 · lost 1
Gemma 7B Itwon 7 · lost 1
Mistral Instruct v0.1 (7B)won 6 · lost 1
DeepSeek LLM 67B Chatwon 7 · lost 2
GPT-3.5 Turbowon 7 · lost 2
Mistral 7B Instruct V0.2won 6 · lost 2
Llama 3.2 3B Instructwon 6 · lost 2
Command Rwon 5 · lost 2
Jamba 1.5 Miniwon 7 · lost 3
Command R Pluswon 7 · lost 4
Qwen1.5 14B Chatwon 5 · lost 3
Magistral Mediumwon 5 · lost 3
Jamba 1.5 Largewon 6 · lost 4
Mistral Smallwon 6 · lost 4
NVIDIA Nemotron 3 Ultra 550B A55Bwon 6 · lost 4
Mistral Instruct v0.3 (7B)won 4 · lost 3
OLMo 2 7B Instructwon 4 · lost 3
Llama 3 8B Instructwon 5 · lost 4
Gemma 2 9B Itwon 5 · lost 4
Llama 3.1 70B Instructwon 6 · lost 5
Llama 3.2 1B Instructwon 4 · lost 4
Phi 3 Mini 128k Instructwon 4 · lost 4
Phi 3 Mini 4k Instructwon 4 · lost 4
Granite 3.0 2B Instructwon 4 · lost 4
Phi 3 Small 8k Instructwon 4 · lost 4
Granite 3.1 2B Instructwon 4 · lost 4
Gemma 2 2B Itwon 4 · lost 4
Granite 3.1 8B Instructwon 4 · lost 4
Mistral Small 24B Instructwon 4 · lost 4
Gemma 3 4B Itwon 4 · lost 4
QwQ 32Bwon 4 · lost 4
NVIDIA Nemotron 3 Super 120B A12Bwon 4 · lost 4
Qwen3 30B A3B Instructwon 4 · lost 4
Command Awon 5 · lost 5
GLM-4.6won 5 · lost 5
Grok 4.1 Fastwon 5 · lost 5
GLM-5.2won 5 · lost 5
Gemini 3.6 Flashwon 5 · lost 5
Llama 3.1 8B Instructwon 5 · lost 6
Grok 3won 5 · lost 6
Gemma 4 31Bwon 4 · lost 5
IBM Granite 4.0 Microwon 3 · lost 4
Gemini 1.5 Flashwon 4 · lost 6
DeepSeek V4 Flashwon 4 · lost 6
Grok 4.3won 4 · lost 6
Mimo V2.6 Flashwon 4 · lost 6
Gemini 3.5 Flashwon 4 · lost 6
Ling 3.0 Flashwon 3 · lost 5
Qwen 2.5 Instruct (7B)won 3 · lost 5
Starling Lm 7B Betawon 3 · lost 5
Granite 3.0 8B Instructwon 3 · lost 5
Phi 3 Medium 4k Instructwon 3 · lost 5
Llama 3.1 Tulu 3 8Bwon 3 · lost 5
C4ai Aya Expanse 8Bwon 3 · lost 5
Nova Prowon 3 · lost 5
Gemma 4 26B A4Bwon 3 · lost 5
GPT OSS 20Bwon 4 · lost 7
Llama 4 Scout Instructwon 4 · lost 7
GPT-4.1 Nanowon 4 · lost 7
Gemini 1.5 Prowon 4 · lost 7
Grok 4won 4 · lost 7
DeepSeek R1won 4 · lost 7
Llama 3.3 70B Instructwon 3 · lost 6
Gemini 2.0 Flash Litewon 3 · lost 6
Mistral Largewon 3 · lost 7
GLM-4.5won 3 · lost 7
DeepSeek V3.2won 3 · lost 7
Kimi K2.5won 3 · lost 7
Kimi K2.6won 3 · lost 7
DeepSeek V4 Prowon 3 · lost 7
Qwen3.7 Maxwon 3 · lost 7
Llama 3.2 Vision (11B)won 2 · lost 5
Llama 3.2 Vision (90B)won 2 · lost 5
Llama 4 Maverick Instructwon 3 · lost 8
GPT OSS 120Bwon 3 · lost 8
Gemini 2.0 Flashwon 3 · lost 8
DeepSeek V3won 3 · lost 8
Claude Haiku 4.5won 3 · lost 8
Kimi K2 Thinkingwon 2 · lost 6
Nova Microwon 2 · lost 6
Nova Litewon 2 · lost 6
GPT-4won 2 · lost 6
Qwen3.5 27Bwon 2 · lost 6
Qwen3 Vl 235B A22B Instructwon 2 · lost 6
Qwen3.5 122B A10Bwon 2 · lost 6
Claude 3 Haikuwon 2 · lost 7
Claude 3.5 Haikuwon 2 · lost 7
DeepSeek V3.1won 2 · lost 7
Mistral Large 4won 2 · lost 8
MiniMax M3won 2 · lost 8
GPT-6 Lunawon 2 · lost 8
GPT-5.6 Lunawon 2 · lost 8
GPT-6 Solwon 2 · lost 8
GLM-5.1won 2 · lost 8
Grok 4.5won 2 · lost 8
GPT-5.6 Terrawon 2 · lost 8
GPT-5.4won 2 · lost 8
GPT-5.5won 2 · lost 8
Qwen3.8 Maxwon 2 · lost 8
GPT-5.6 Solwon 2 · lost 8
Gemini 3.8 Flashwon 2 · lost 8
GPT-4o Miniwon 2 · lost 9
Llama 3.1 405B Instructwon 2 · lost 9
GPT-5 Nanowon 2 · lost 9
GPT-4.1 Miniwon 2 · lost 9
Gemini 2.5 Flashwon 2 · lost 9
Kimi K2won 2 · lost 9
Qwen3 235B A22B Instructwon 2 · lost 9
Gemini 2.5 Prowon 2 · lost 9
Phi 4won 1 · lost 7
C4ai Aya Expanse 32Bwon 1 · lost 7
Qwen Maxwon 1 · lost 7
Gemma 3 12B Itwon 1 · lost 7
Mercury 2won 1 · lost 7
Gemma 3 27B Itwon 1 · lost 7
Qwen3.5 35B A3Bwon 1 · lost 7
Qwen3.5 397B A17Bwon 1 · lost 7
Hy3won 1 · lost 7
Claude Opus 4.6won 1 · lost 7
Llama 3 70B Instructwon 1 · lost 8
Claude 3 Sonnetwon 1 · lost 8
Claude 3 Opuswon 1 · lost 8
GPT-5.2won 1 · lost 9
Grok 4.6won 1 · lost 9
Mimo V2.5 Prowon 1 · lost 9
GLM-5.3 Flashwon 1 · lost 9
Mimo V2.6 Prowon 1 · lost 9
Claude Opus 5won 1 · lost 9
Muse Spark 1.1won 1 · lost 9
Claude Opus 5.5won 1 · lost 9
O1 Miniwon 1 · lost 10
O3 Miniwon 1 · lost 10
Claude 3.7 Sonnetwon 1 · lost 10
GPT-5 Miniwon 1 · lost 10
O4 Miniwon 1 · lost 10
GPT-4.1won 1 · lost 10
O3won 1 · lost 10
GPT-5won 1 · lost 10
GPT-5.1won 1 · lost 10
Claude Sonnet 4.5won 1 · lost 10
MiniMax M2won 0 · lost 8
Grok 3 Miniwon 0 · lost 8
Gemini 2.5 Flash Litewon 0 · lost 8
Claude Sonnet 4.6won 0 · lost 8
Gemini 3 Prowon 0 · lost 8
Qwen2 72B Instructwon 0 · lost 9
Gemma 2 27B Itwon 0 · lost 9
MiniMax M2.7won 0 · lost 10
Inklingwon 0 · lost 10
Grok 4.7won 0 · lost 10
Claude Sonnet 5won 0 · lost 10
Claude Opus 4.8won 0 · lost 10
Claude Opus 4.7won 0 · lost 10
Qwen2.5 72B Instructwon 0 · lost 11
GPT-4 Turbowon 0 · lost 11
GPT-4owon 0 · lost 11
Claude 3.5 Sonnetwon 0 · lost 11
O1won 0 · lost 11
Claude Sonnet 4won 0 · lost 11
Claude Opus 4won 0 · lost 11

§ 3 · Sources

Where the numbers come from

4 publications, 13 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-10-19
LMArena Text 1230
Vals.ai ↗1 measureread 2026-10-19
Vals · MedQA 62.14%
HELM Safety ↗5 measuresread 2026-10-19
HELM Safety · HarmBench 52.1%HELM Safety · SimpleSafetyTests 91.5%HELM Safety · Anthropic Red Team 96.2%HELM Safety · BBQ 91.6%HELM Safety · XSTest 94.1%
Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-10-19
Enkrypt · Jailbreak risk 8%Enkrypt · Harmful content risk 46.1%Enkrypt · CBRN risk 9.3%Enkrypt · Toxicity risk 2.6%Enkrypt · Bias risk 88.1%Enkrypt · Insecure code risk 40.9%

Badge

PublicAI Index badge for Mixtral 8x22b Instruct[![PublicAI Index](https://publicai.io/model-index/badge?model=mixtral-8x22b-instruct)](https://publicai.io/model-index/m/mixtral-8x22b-instruct)