Mistral Large 4Open Weight

Released

Mistral Large 4 is an AI model released by Mistral on Tuesday, Oct 6 2026, 161 days after Mistral Medium 3.5. It is an open-weight model — the trained weights are available to download and run. It is a 1.05T parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 1M token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover SWEAtlas CodeBase QnA, BullshitBench v2, DeepSWE 1.1, FrontierCode 2.0, AA Coding Agent Index, Terminal-Bench 4.0, and 4 more.

API pricing

Input
$0.70
Cached input
$0.07
Output
$2.10
  • Reasoning or thinking is supported.
USD per 1M tokensVerified October 19, 2026Read from mistral.ai. Click the check to open the source.

Available from

Mistral—zdr$0.70$2.101M
Mistraleu—$0.77$2.31512K
USD per 1M tokensChecked November 20, 2026Third-party rates from openrouter.ai, not the lab’s own list price.

Benchmarks

Coding

SWEAtlas CodeBase QnACodebase understanding — Questions about how an unfamiliar codebase actually works — where something is handled, what a change would touch — answered by reading the repository rather than editing it. Tests understanding rather than patch-writing. Higher is better.
59.4%
#1
DeepSWE 1.1Agentic coding — Artificial Analysis' independent test of deep, agentic software-engineering work — the AI has to plan and carry out substantial coding tasks end to end. (Version 1.1 of the test.) Higher is better.
61.7%
#28 of 43
FrontierCode 2.0Agentic coding — whether a model’s code changes to open-source repositories would be merged by their maintainers (version 2.0); separate from FrontierCode v1.1 scores
19.7%
#9 of 9
AA Coding Agent IndexAgentic coding — Artificial Analysis' overall score for coding agents, combining three coding benchmarks with what each run costs and how many tokens it burns. It rates a model paired with a particular agent harness rather than the model alone, so the same model scores differently in different tools. Higher is better.
48
#2 of 3

Terminal & CLI

Terminal-Bench 4.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Version 4.0 recalibrated how much time, CPU and memory each task gets, removed eight tasks and fixed nineteen, so fewer runs fail for reasons that have nothing to do with the model. Scores are not comparable with earlier versions. Higher is better.
41%
#21 of 44

Agentic & tool use

BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better.
84.5%
#13 of 48
AutomationBenchBusiness workflows — Tests whether the AI can run real multi-step business workflows — the kind of end-to-end office processes companies want to automate — from start to finish. Higher is better.
59.9%
#2 of 33

Knowledge work

AA-Briefcase v1.1Knowledge work — Artificial Analysis agentic office-work eval (Elo, v1.1 rating fit)
1393
#13 of 15

Robustness

BullshitBench v2Nonsense detection — Given a confidently-worded but nonsensical prompt, does the AI spot that it makes no sense and push back — instead of playing along and inventing an answer? The score is how often it clearly called out the nonsense. Higher is better.via BullshitBench
80%
#7 of 107

Community preference

threejsevalCommunity preference (Three.js) — Every model gets the same prompt — "the Eiffel Tower", "a glass fishbowl", "a robot arm picking toys into a box" — and builds a 3D scene in Three.js. Real people then see two scenes side by side, names hidden, and vote for the one they prefer. The votes become a chess-style Elo rating on threejseval.com, averaged across all the prompts. It measures whether the scene looks and moves right to a human eye, not whether the code passes a test. Higher is better.via threejseval
1008
#31 of 31

Compare Mistral Large 4 with

Mistral Large 4

About

Mistral Large 4 entered public API preview in October 2026 as a natively multimodal mixture-of-experts model for reasoning, coding and agent tasks. The preview was available through Mistral’s API and Mistral Studio. Mistral described training it from scratch on NVIDIA Grace Blackwell hardware in its European datacenters.

Mistral released the weights on October 29, 2026, with a model improved by three more weeks of reinforcement learning since the preview. The release post reported 41% on Terminal-Bench 4.0, up from 28% in the preview, 48% on the Artificial Analysis Coding Agent Index with fewer tokens per task than GLM-5.3, and 84.5% on BrowseComp. It added low, medium and high reasoning modes and cyber guardrails meant to separate defensive work from attacks, while a specialized cybersecurity variant went only to vetted design partners. The weights shipped under the Mistral Ecosystem Open-Weight license, a modified MIT license that requires a separate commercial license for model-as-a-service providers. These were vendor-reported results under Mistral's own setups.

Frequently asked questions

Mistral Large 4 was released by Mistral on Tuesday, Oct 6 2026.

All Mistral releases

45 tracked