Start a Conversation With Mistral Large 4
Pick a ready-made task or type your own prompt — the API call is handled for you, so there is no key to paste and no code to write
None

None

None

Long Story Video Skill

Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.

Ads Video Skill

Ads Video Skill

Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.

3D Science Explainer Video Skill

Convert scientific concepts into stunning 3D explain animations

AI Video Prompt Generator

Feedback

freeTrialImage.bannerPity

freeTrialImage.upgradeUnlock

  • ✓freeTrialImage.benefitHd
  • ✓freeTrialImage.benefitWatermark
  • ✓freeTrialImage.benefitUnlimited

Mistral Large 4

Put Mistral Large 4 to work on code, contracts and charts. Fifteen credits refresh daily, sign-in takes seconds, and no API key is ever needed.

All Tools

Discover our comprehensive AI-powered animation toolkit

Inside This Mistral Large 4 Workspace

One tab covers a 1.05-trillion-parameter mixture-of-experts: sort security logs, read contracts, check financial models or ask about a screenshot — with credits renewed daily and nothing to install.

  • Trillion-Parameter Scale With Built-In Sight
    Public preview has run since October 6, 2026. The 1.05-trillion-parameter MoE pairs with a 1.6B vision encoder, so an image and a question can arrive in the same prompt.
  • Fifteen Credits, Renewed Daily
    Google or email sign-in takes seconds; 15 credits appear at midnight UTC, and neither a key nor any code is required.
  • Frontier Scores at a Fraction of the Cost
    Launch-sale pricing lands near $1.03 per million blended tokens — roughly one-nineteenth of GPT-6 Astra and one-eighth of Claude Opus 5.5.

Three Steps to a Real Mistral Large 4 Answer

From a blank page to a usable reply in three moves.

Where Mistral Large 4 Pulls Ahead

Six strengths, drawn from the benchmarks Mistral put front and centre at launch.

Vulnerability Hunting and Security Review

Its strongest showing: 93% on Cybench, 82% on CyberGym-E2E and 50 tasks solved on the AA Cyber Index — useful for code review, log triage and writing up a fix.

Contracts, Clauses and Financial Models

Vals AI's Harvey Legal Agent Benchmark scores it 15.8 — sixth of 75 models and nearly triple GPT-6 Astra's 5.4 — while Finance Agent v2 returns 54.7.

Object Grounding and Chart Comprehension

Dense200 reaches 42.0 for locating objects in crowded scenes, and ChartQA Pro 63.1 for chart reading. Upload a photo and ask for bounding boxes.

Agentic Coding With Structured JSON

DeepSWE v1.1 lands at 61.7%, and the Coding Agent Index puts it above DeepSeek V4 Pro and Qwen3.8 Max, with function calling, structured JSON and reasoning_effort: "high" supported out of the box.

Coverage Across 160+ Languages

Training spanned more than 160 languages, every official EU language among them, so drafting, translation and support replies all stay inside one model.

Fast Output at 116.1 Tokens a Second

Artificial Analysis measured 116.1 tokens/s against a class median of 87, alongside a $1.03 blended rate per 1M tokens during the launch sale.

FAQ

Mistral Large 4: Your Questions Answered

Clear answers on access, credits, plans, privacy and the upcoming open weights.

1

Do I need a Mistral API key to use this chat?

No. Your prompt is forwarded to the Mistral API on your behalf, so there is no key, billing account or code involved — just open the page and start typing.

2

Why do free replies move to Mistral Small 4 after three messages?

Free accounts begin on Mistral Large 4 for three replies, then continue on Mistral Small 4. Pro plans and credit packs keep every answer on the larger model.

3

How are credits counted for each reply?

You pay for the tokens actually used: your message, the context sent with it and the answer. A short exchange costs 1 credit, the exact amount appears under each reply, and an interrupted answer bills only the text delivered.

4

Where is my chat history stored, and can I pick it up on another device?

Conversations are tied to your account on every plan, so you can reopen them later from a phone, tablet or laptop.

5

Is this the official Mistral website?

No — it is an independent front-end that calls the Mistral API. Benchmark and pricing figures come from Mistral's launch announcement and provider pricing pages checked on October 6, 2026.

6

When do the open weights for Mistral Large 4 ship?

Mistral is targeting the end of October 2026: the Hugging Face model page lists October 31, while some press reports point to October 27.

Give Mistral Large 4 a Real Task

Paste a broken function, a contract clause, a rough spreadsheet note or a screenshot. Free accounts get 15 credits every day, and there is nothing to install or configure.