Mistral: Mistral Large 3 2512
by Mistral AItext+image+file->text2 endpoints
Neura Intelligence Index
54.9/ 100
Rank
#173—
Confidence
highBenchmarks
10(26% coverage)
Overview
Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.
Capabilities
Text generation
Image understanding
Tool use / Function calling
Structured output
Prompt caching
Long context
File input
Top-P sampling
Stop sequences
Deterministic seed
Modalities
Input
TextImageFile
Output
Text
Technical Specifications
Context Window
262.1K tokens
Max Output
209.7K tokens
Tokenizer
Mistral
Uptime (24h)
10000.0%
Parameters
123B
Blended Cost
$3/1M tokens
License
mistral-researchOpen Weights
Country
🇫🇷FR
Model Family
Mistral Large
Release Date
July 24, 20242y
Supported Parameters
frequency_penaltymax_tokenspresence_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_p
Benchmark Performance
general
| Benchmark | Score | Source |
|---|---|---|
| Chatbot Arena Elo | 1250 | aggregated |
| IFEval | 82.8 | aggregated |
| MMLU-Pro | 69.4 | aggregated |
coding
| Benchmark | Score | Source |
|---|---|---|
| HumanEval | 92 | aggregated |
reasoning
| Benchmark | Score | Source |
|---|---|---|
| ARC-Challenge | 94 | aggregated |
| GPQA Diamond | 48.2 | aggregated |
language
| Benchmark | Score | Source |
|---|---|---|
| HellaSwag | 93.6 | aggregated |
safety
| Benchmark | Score | Source |
|---|---|---|
| TruthfulQA | 60.5 | aggregated |
Performance Over Time
Pricing
Per 1M tokens
Input$0.50
Output$1.50
Cache read$0.05
Added
December 1, 2025
Last synced 10/23/2026