Mistral Large 3

Creator

Mistral AI

Released

2025-12-02

Intelligence

15.9

Artificial Analysis Index

Coding

20.1

Artificial Analysis Index

In $/1M

$0.50

input tokens

Out $/1M

$1.50

output tokens

Blended $/1M

$0.75

3:1 blended

Speed

62

tokens / sec

Profile

License, openness, and modality — the governance layer.

License

Apache 2.0

commercial OK

Weights

Open

downloadable

Modalities

Image · Text

Parameters

675B

total

Mistral Large 3 is a state-of-the-art general-purpose Multimodal granular Mixture-of-Experts model with 41B active parameters and 675B total parameters trained from scratch with 3000 H200s. This model is the instruct post-trained version, fine-tuned for instruction tasks, making it ideal for chat, agentic and instruction based use cases. Designed for reliability and long-context comprehension - It is engineered for production-grade assistants, retrieval-augmented systems, scientific workloads, and complex enterprise workflows.

Capability profile

Category strength across 20 domains, via LLM Stats.

multimodal
100communication
100general
100math
80
language
80reasoning
70writing
60creativity
60
Via LLM Stats

Benchmark breakdown

Independent evaluation scores, normalized to 0–100.

GPQA Diamond
68
Humanity’s Last Exam
4
MMLU-Pro
81
SciCode
36
LiveCodeBench
47
MATH-500
AIME 2025
38
τ²-Bench (agentic)
25
Terminal-Bench Hard
16
IFBench
36
Via Artificial Analysis
© 2026 NYSGPT2525 LLC