The LLM benchmark aggregator.

Mistral Small 3

Mistral · 24B · open weights

Strongest in Safe-prompt compliance (#50 of 82), weakest in Harm refusal (#269 of 299). Above par in 2 of 4 scopes. Among the models it meets almost everywhere, it finishes behind Claude 3.5 Sonnet and GPT-5 Nano and ahead of Qwen2.5 3B Instruct and Jamba 1.5 Mini.

§ 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.

Safety47.4−14.1#243/3361/3

Safe-prompt compliance52.6−7.8#50/821/1

Fairness53.9−16.1#77/2991/2

Harm refusal42.9−18.7#269/2991/2

§ 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 Mistral Small 3, left for the other.

OLMo 7B Instruct Hfwon 3 · lost 0

RakutenAI 7B Chatwon 3 · lost 0

H2o Danube3 500M Chatwon 3 · lost 0

LFM2 350Mwon 3 · lost 0

Qwen3 30B A3B Instructwon 3 · lost 0

Laguna Xs 2.1won 3 · lost 0

Nous Hermes 2 Mixtral 8x7b Dpowon 3 · lost 0

Qwen2.5 0.5B Instructwon 3 · lost 0

Jamba Mini 1.7won 3 · lost 0

Laguna S 2.1won 3 · lost 0

Falcon H1 0.5B Instructwon 3 · lost 0

Magistral Mediumwon 3 · lost 0

Command R7Bwon 3 · lost 0

Command Awon 3 · lost 0

Jamba Large 1.7won 3 · lost 0

Qwen3 8Bwon 3 · lost 0

OLMoE 1B-7B Instructwon 4 · lost 0

Mistral Instruct v0.1 (7B)won 4 · lost 0

DBRX Instructwon 4 · lost 0

OLMo 2 13B Instructwon 4 · lost 0

Mistral Instruct v0.3 (7B)won 4 · lost 0

Command Rwon 4 · lost 0

Mixtral (8x7B)won 4 · lost 0

OLMo 2 7B Instructwon 4 · lost 0

Grok 4won 3 · lost 1

GPT-3.5 Turbowon 3 · lost 1

Llama 3.1 Instruct Turbo (70B)won 3 · lost 1

Command R Pluswon 3 · lost 1

DeepSeek LLM 67B Chatwon 3 · lost 1

Marin 8B Instructwon 3 · lost 1

Mixtral Instruct (8x22B)won 3 · lost 1

IBM Granite 4.0 Micro (with guardian)won 3 · lost 1

SmolLM 1.7B Instructwon 2 · lost 1

PowerMoE 3Bwon 2 · lost 1

Phi 2won 2 · lost 1

Qwen2.5 7B Instructwon 2 · lost 1

Gemma 7B Itwon 2 · lost 1

PowerLM 3Bwon 2 · lost 1

Smollm2 1.7B Instructwon 2 · lost 1

Grok 4.1 Fastwon 2 · lost 1

RakutenAI 7B Instructwon 2 · lost 1

Falcon Mamba 7B Instructwon 2 · lost 1

GLM-4 9B Chatwon 2 · lost 1

CodeLlama 7B Instruct Hfwon 2 · lost 1

Qwen2 7B Instructwon 2 · lost 1

Gemma 2B Itwon 2 · lost 1

Magistral Smallwon 2 · lost 1

Aya 23 8Bwon 2 · lost 1

SEA LION V1 7B ITwon 2 · lost 1

Llama 3.1 8B Instructwon 2 · lost 1

Grok Build 0.1won 2 · lost 1

Qwen1.5 14B Chatwon 2 · lost 1

DeepSeek R1 Distill Qwen 7Bwon 2 · lost 1

DeepSeek R1 Distill Llama 8Bwon 2 · lost 1

H2o Danube3 4B Chatwon 2 · lost 1

Zephyr 7B Betawon 2 · lost 1

Ling 3.0 Flash VLwon 2 · lost 1

Mimo V2.6 Flashwon 2 · lost 1

Gemini 3.6 Flashwon 2 · lost 1

Jamba 1.6 Miniwon 2 · lost 1

LFM2 700Mwon 2 · lost 1

DeepSeek V4 Flashwon 2 · lost 1

NexusRaven V2 13Bwon 2 · lost 1

Ling 3.0 Flashwon 2 · lost 1

Open Mixtral 8x7bwon 2 · lost 1

Jamba 1.5 Largewon 2 · lost 1

Gemma 2 9B Itwon 2 · lost 1

GLM-5.2won 2 · lost 1

Gemini 3.5 Flashwon 2 · lost 1

Granite 3.2 8B Instructwon 2 · lost 1

Mistral 7B Instruct V0.2won 2 · lost 1

Nemotron 3 Ultra 550B A55Bwon 2 · lost 1

GLM-4.6won 2 · lost 1

Grok 4.3won 2 · lost 1

Phi-4 Mini Instructwon 2 · lost 1

Open Codestral Mambawon 2 · lost 1

Mimo V2.6 Pro Ultraspeedwon 2 · lost 1

Jamba 1.5 Miniwon 2 · lost 1

Qwen2.5 3B Instructwon 2 · lost 1

Grok 3won 2 · lost 2

DeepSeek R1won 2 · lost 2

DeepSeek V3won 2 · lost 2

IBM Granite 4.0 Microwon 2 · lost 2

Gemini 2.0 Flash Litewon 2 · lost 2

IBM Granite 4.0 Small (with guardian)won 2 · lost 2

IBM Granite 3.3 8B Instructwon 2 · lost 2

Gemini 1.5 Prowon 2 · lost 2

Gemini 1.5 Flashwon 2 · lost 2

Claude 3 Haikuwon 2 · lost 2

IBM Granite 4.0 Smallwon 2 · lost 2

GPT-4.5won 2 · lost 2

O3 Miniwon 2 · lost 2

Claude 3 Sonnetwon 2 · lost 2

SmolLM2 360M Instructwon 1 · lost 2

SmolLM 360M Instructwon 1 · lost 2

LFM2 1.2Bwon 1 · lost 2

Aya 23 35Bwon 1 · lost 2

NVIDIA Nemotron 3 Super 120B A12Bwon 1 · lost 2

MiniMax M3won 1 · lost 2

Granite 3.0 1B A400M Instructwon 1 · lost 2

DeepSeek V3p1won 1 · lost 2

Gemma 4 E2Bwon 1 · lost 2

QwQ 32Bwon 1 · lost 2

Jamba 1.6 Largewon 1 · lost 2

Yi 1.5 9B Chatwon 1 · lost 2

Gemma 4 E4Bwon 1 · lost 2

Jamba Instructwon 1 · lost 2

Qwen3.7 Maxwon 1 · lost 2

SeaLLMs V3 7B Chatwon 1 · lost 2

Qwen2.5 7B Instruct 1Mwon 1 · lost 2

Mistral Smallwon 1 · lost 2

Kimi K2.5won 1 · lost 2

DeepSeek Reasonerwon 1 · lost 2

Qwen2.5 1.5B Instructwon 1 · lost 2

Mistral Small 24B Instructwon 1 · lost 2

DeepSeek V3.2won 1 · lost 2

Granite 3.1 2B Instructwon 1 · lost 2

WizardLM 2.8x22Bwon 1 · lost 2

Llama 3.2 3B Instructwon 1 · lost 2

Gemma 2 2B Itwon 1 · lost 2

Kimi K2.7 Codewon 1 · lost 2

Gemma 4 12Bwon 1 · lost 2

Hunyuan A13B Instructwon 1 · lost 2

Granite 3.2 2B Instructwon 1 · lost 2

Fuguwon 1 · lost 2

Llama 3.3 Instruct Turbo (70B)won 1 · lost 2

Granite 3.0 3B A800M Instructwon 1 · lost 2

Gemma 3 4B Itwon 1 · lost 2

Gemma 4 31Bwon 1 · lost 2

Starling Lm 7B Betawon 1 · lost 2

Open Mixtral 8x22bwon 1 · lost 2

Granite 3.0 2B Instructwon 1 · lost 2

Gemma2 9B Cpt Sea Lionv3 Instructwon 1 · lost 2

Reflection Llama 3.1 70Bwon 1 · lost 2

Apertus 8B Instructwon 1 · lost 2

Llama 3.2 1B Instructwon 1 · lost 2

Gemma 4 26B A4Bwon 1 · lost 2

DeepSeek Chatwon 1 · lost 2

DeepSeek V4 Prowon 1 · lost 2

Qwen2.5 32B Instructwon 1 · lost 2

LongWriter Glm4 9Bwon 1 · lost 2

Qwen2.5 14B Instruct 1Mwon 1 · lost 2

Qwen2 VL 72Bwon 1 · lost 2

GLM-5 Turbowon 1 · lost 2

Granite 3.1 3B A800M Instructwon 1 · lost 2

Phi 3 Medium 4k Instructwon 1 · lost 2

Granite 3.1 8B Instructwon 1 · lost 2

Sarvam 30Bwon 1 · lost 2

Qwen2.5 14B Instructwon 1 · lost 2

Qwen3 Vl 235B A22B Instructwon 1 · lost 2

Phi 3 Medium 128k Instructwon 1 · lost 2

Gemma 3 27B Itwon 1 · lost 2

Qwen2 57B A14B Instructwon 1 · lost 2

Smaug Llama 3 70B Instructwon 1 · lost 2

Llama 3 70B Chat HFwon 1 · lost 2

Phi 3 Small 128k Instructwon 1 · lost 2

Granite 3.0 8B Instructwon 1 · lost 2

Claude Opus 4.6won 1 · lost 2

Claude Sonnet 4.6won 1 · lost 2

Mimo V2.6 Prowon 1 · lost 2

GPT OSS Safeguard 20Bwon 1 · lost 2

Gemma 3 12B Itwon 1 · lost 2

SeaLLM 7B V2won 1 · lost 2

Apertus 70B Instructwon 1 · lost 2

Granite 4.0 H Microwon 1 · lost 2

Granite 4.0 H Tinywon 1 · lost 2

Phi 3 Small 8k Instructwon 1 · lost 2

Phi 3 Mini 128k Instructwon 1 · lost 2

Phi 3.5 Mini Instructwon 1 · lost 2

K2 Chatwon 1 · lost 2

Llama 3.1 Tulu 3 8B SFTwon 1 · lost 2

Phi 3 Mini 4k Instructwon 1 · lost 2

Nova Prowon 1 · lost 2

GPT-5.6 Solwon 1 · lost 2

Qwen3.8 Maxwon 1 · lost 2

Llama 3.1 Tulu 3 8B DPOwon 1 · lost 2

Krutrim 2 Instructwon 1 · lost 2

GLM-5.1won 1 · lost 2

Llama 3.1 Tulu 3 8Bwon 1 · lost 2

Qwen3.5 35B A3Bwon 1 · lost 2

EAI SageAlign 8B LoRAwon 1 · lost 2

Qwen Maxwon 1 · lost 2

GPT-4won 1 · lost 2

Phi 3.5 MoE Instructwon 1 · lost 2

Nova Microwon 1 · lost 2

Llama 3 8B Chat Hfwon 1 · lost 2

GPT-5.6 Lunawon 1 · lost 2

Nova Litewon 1 · lost 2

C4ai Aya Expanse 32Bwon 1 · lost 2

C4ai Aya Expanse 8Bwon 1 · lost 2

Llama 3.2 Vision (11B)won 1 · lost 2

Titan Text Express V1won 1 · lost 2

Internlm2 Chat 20Bwon 1 · lost 2

Qwen2.5 72B Instructwon 1 · lost 2

Llama 2 70B Chat Hfwon 1 · lost 2

Llama 2 13B Chat Hfwon 1 · lost 2

Llama 3 8B Instruct RRwon 1 · lost 2

Nova Premierwon 1 · lost 2

Granite 4.0 H Smallwon 1 · lost 2

Llama 2 7B Chat Hfwon 1 · lost 2

Sarvam 105Bwon 1 · lost 2

GPT-5.4won 1 · lost 2

Llama 3 8B Instruct MopeyMulewon 1 · lost 2

Cogito 671B V2 P1won 1 · lost 2

Claude Opus 4.7won 1 · lost 2

Meta SecAlign 8Bwon 1 · lost 2

Apriel 1.5 15B Thinkerwon 1 · lost 2

Phi 4won 1 · lost 2

Palmyra Medwon 1 · lost 3

Llama 4 Scout Instructwon 1 · lost 3

DeepSeek R1 (hide reasoning)won 1 · lost 3

Llama 3.1 Instruct Turbo (8B)won 1 · lost 3

Olmo 2 32B Instructwon 1 · lost 3

Llama 3.1 Instruct Turbo (405B)won 1 · lost 3

Qwen1.5 72B Chatwon 1 · lost 3

GPT OSS 20Bwon 1 · lost 3

Llama 3 8B Instructwon 1 · lost 3

Gemini 2.0 Flashwon 1 · lost 3

Llama 4 Maverick Instructwon 1 · lost 3

Llama 3 70B Instructwon 1 · lost 3

GPT-4.1 Nanowon 1 · lost 3

GPT-4o Miniwon 1 · lost 3

Palmyra X5won 1 · lost 3

GPT-4.1 Miniwon 1 · lost 3

O4 Miniwon 1 · lost 3

Qwen3 235B A22B Instructwon 1 · lost 3

GPT OSS 120Bwon 1 · lost 3

Qwen3 Next 80B A3B Thinkingwon 1 · lost 3

Claude 3.7 Sonnetwon 1 · lost 3

GPT-4owon 1 · lost 3

Qwen2 72B Instructwon 1 · lost 3

Claude Haiku 4.5won 1 · lost 3

GPT-4.1won 1 · lost 3

O1 Miniwon 1 · lost 3

O1won 1 · lost 3

Claude 3 Opuswon 1 · lost 3

GLM-4.5won 0 · lost 3

Kimi K2.6won 0 · lost 3

Sarvam 1won 0 · lost 3

Komodo 7B Basewon 0 · lost 3

Mimo V2.5 Prowon 0 · lost 3

Dolphin 2.5 Mixtral 8x7bwon 0 · lost 3

GLM-5.3 Flashwon 0 · lost 3

GPT-5.6 Astrawon 0 · lost 3

Kimi K2 Thinkingwon 0 · lost 3

Gemini 3.8 Flashwon 0 · lost 3

EAI Llama 3.1 8B Instruct SageAlignwon 0 · lost 3

GPT-6 Solwon 0 · lost 3

Mistral Largewon 0 · lost 3

Qwen3.5 27Bwon 0 · lost 3

GPT-6 Lunawon 0 · lost 3

Qwen3.5 122B A10Bwon 0 · lost 3

Llama 3.1 Nemotron 70B Instruct HFwon 0 · lost 3

Fugu Ultra V2won 0 · lost 3

Grok 4.5won 0 · lost 3

GPT-5.5won 0 · lost 3

Llama 3.2 90B Vision Instruct Turbowon 0 · lost 3

GPT-5.6 Terrawon 0 · lost 3

Foundation Sec 8B Instructwon 0 · lost 3

Gemma 2 27B Itwon 0 · lost 3

Qwen3.5 397B A17Bwon 0 · lost 3

Mistral NeMo Minitron 8B Instructwon 0 · lost 3

Claude Opus 5won 0 · lost 3

GPT-5.2won 0 · lost 3

Grok 4.6won 0 · lost 3

MiniMax M2won 0 · lost 3

Claude 3.5 Haikuwon 0 · lost 3

Muse Spark 1.1won 0 · lost 3

Claude Opus 4.8won 0 · lost 3

Inklingwon 0 · lost 3

Grok 4.7won 0 · lost 3

Hy3won 0 · lost 3

MiniMax M2.7won 0 · lost 3

Claude Opus 5.5won 0 · lost 3

Claude Sonnet 5won 0 · lost 3

Gemini 2.5 Prowon 0 · lost 4

Gemini 2.5 Flashwon 0 · lost 4

Gemini 3 Prowon 0 · lost 4

Qwen 2.5 Instruct Turbo (7B)won 0 · lost 4

Gemini 2.0 Prowon 0 · lost 4

Kimi K2won 0 · lost 4

Qwen3 235B A22B Throughputwon 0 · lost 4

GLM-4.5 Airwon 0 · lost 4

Grok 3 Miniwon 0 · lost 4

Palmyra Xwon 0 · lost 4

Palmyra Finwon 0 · lost 4

Gemini 2.5 Flash Litewon 0 · lost 4

Claude Opus 4won 0 · lost 4

Qwen 2.5 Instruct Turbo (72B)won 0 · lost 4

Claude Sonnet 4.5won 0 · lost 4

Claude Sonnet 4won 0 · lost 4

GPT-4 Turbowon 0 · lost 4

GPT-5.1won 0 · lost 4

O3won 0 · lost 4

GPT-5 Miniwon 0 · lost 4

GPT-5won 0 · lost 4

GPT-5 Nanowon 0 · lost 4

Claude 3.5 Sonnetwon 0 · lost 4

§ 3 · Sources

Where the numbers come from

1 publication, 5 figures. Every one links to the page it was read from.

HELM Safety ↗5 measuresread 2026-09-27

HELM Safety · HarmBench 45.2%HELM Safety · SimpleSafetyTests 93.3%HELM Safety · Anthropic Red Team 96.4%HELM Safety · BBQ 93.3%HELM Safety · XSTest 95.3%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=mistral-small-3)](https://publicai.io/model-index/m/mistral-small-3)