Mistral Large 3

Available

A Mistral AI model with a 256K-token context window, image input and tool calling. On the Mistral API since December 2, 2025.

Access details: Mistral API

The model ID to use in API requests.

mistral-large-2512

Updated Sep 23, 2026

Sources & notes

Capabilities & limits

Context window: 256K
Maximum output: 256K
Image input: Supported
Tool calling: Supported
Structured output: Supported

Pricing

USD / 1M tokens

Input: $0.50
Output: $1.50

Standard rates per million tokens. Discounts and long-context rates, where offered, are on Mistral AI's pricing page.

Full rate card 3 rates

All prices in USD.

Standard input

$0.50 per 1M tokens

Cached input

$0.05 per 1M tokens

Standard output

$1.50 per 1M tokens

Batch processing is billed at a 50% discount. Regional inference is billed at 1.1x standard list pricing for input, output, cached reads and cache writes. Priority Tier is an entitlement; the docs pricing page's Standard tab is recorded here.

Price history

Sep 13, 2026

Input $0.50 per million tokens

Sep 23, 2026

Input $0.50 per million tokens

Sep 13, 2026

Output $1.50 per million tokens

Sep 23, 2026

Output $1.50 per million tokens

Technical details

Parameters: 675B
Active parameters: 41B
Architecture: sparse mixture-of-experts
Training / knowledge cutoff: July 2025
Input modalities: Text, image
Output modalities: Text

From Mistral AI

Official announcement ↗

How they describe it

Open-weight, general-purpose, flagship multimodal and multilingual model.

Release & lifecycle

First available: Dec 2, 2025
available: As of Sep 13, 2026
available: As of Oct 20, 2026