Model

Mistral Large

CLOSED

Mistral AI · Mistral Large family · released Feb 26, 2024

Mistral's first GPT-4-class closed flagship, launched with the Microsoft/Azure partnership and Le Chat; retired from the API by 2025.

ReasoningCodingVisionFunction callingTool useAgentic

713.0

Elo · rank #263

Parameters

Undisclosed

Active params

—

Context

32K tokens

Architecture

Dense transformer (size undisclosed)

License

Proprietary

Languages

API price (in/out)

No hosted API

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

ARC-ChallengeReasoning94.0%#13

best: Llama 3.1 405B · 96.9%

GSM8KMath81.0%#78

best: Llama 3.1 405B · 96.8%

HellaSwagReasoning89.2%#9

best: Claude 3 Opus · 95.4%

HumanEvalCoding45.1%#130

best: Claude Opus 4.5 · 99.4%

MATH-500Math45.0%#141

best: GPT-5 · 99.4%

MBPPCoding73.1%#29

best: Llama-3.3-Nemotron-Super-49B v1 (Reasoning On) · 91.3%

MMLUKnowledge81.2%#80

best: OpenAI o3 · 92.9%

TriviaQAKnowledge82.7%#10

best: Sarvam-1 (2B) · 90.6%

TruthfulQAKnowledge50.5%#53

best: Phi-3.5-MoE (16x3.8B, 6.6B active) · 77.5%

WinoGrandeReasoning86.7%#8

best: PaLM 2 · 90.9%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

—

Output / M tok

—

Mistral Large family

Elo progression across releases

Mistral LargeFeb 2024713.0 Mistral Large 2 (24.07)Jul 20241068.2 Mistral Large 24.11Nov 2024658.0 Mistral Large 3Dec 20251303.2

Resources

DOCSmistral.ai/news/mistral-large

API price — · each benchmark row carries its own source badge (see methodology)