Chat with Mistral Large 4, free, right in your browser
Mistral Large 4 is Mistral AI’s 1.05-trillion-parameter Mixture-of-Experts model, released in public preview on October 6, 2026, with open weights due by the end of October. Its strongest results are in security, legal and finance documents, and finding objects in images. Chat with it here: sign up free with Google or email and get 15 credits every day. No API key, no setup.
- Free account in seconds
- 15 free credits every day
- No API key or setup
Popular prompts
Open chatOpens full-screen chat. Your chats are saved to your account.
- parameters
- 1.05T
- parameters
- 49B active per token
- token context
- 524K
- token context
- 262K max output
- per 1M input tokens
- $0.68
- per 1M input tokens
- 50% launch sale
- 2026 release
- Oct 6
- 2026 release
- open weights due end of Oct
Benchmarks
Where Mistral Large 4 leads
Mistral Large 4 posts its best results on security, legal and visual grounding work. These are scores from Mistral’s launch charts, next to the closest rivals in each test.
Source: Mistral AI launch announcement (Oct 6, 2026) and Artificial Analysis model page, checked Oct 6, 2026.
Real answers
Watch Mistral Large 4 answer
These are real replies from Mistral Large 4, sent through the chat on this page and shown word for word. Pick a task, then ask your own version.
Write a short JavaScript function that groups an array of orders by customer and returns the top 3 customers by total spend. Use camelCase names and add a one-line usage example.
function getTopCustomersBySpend(orders) {
const totalsByCustomer = {};
for (const order of orders) {
totalsByCustomer[order.customerId] =
(totalsByCustomer[order.customerId] || 0) + order.amount;
}
return Object.entries(totalsByCustomer)
.map(([customerId, totalSpend]) => ({ customerId, totalSpend }))
.sort((a, b) => b.totalSpend - a.totalSpend)
.slice(0, 3);
}
// Usage: getTopCustomersBySpend([{customerId: "c1", amount: 100}, {customerId: "c2", amount: 200}, {customerId: "c1", amount: 50}])
// → [{customerId: "c2", totalSpend: 200}, {customerId: "c1", totalSpend: 150}]
Recorded on MistralLarge4.org, 2026-10-07. Reply text unedited.
API price
Frontier quality at a fraction of the price
Blended price per 1M tokens (3 parts input, 1 part output). At the launch sale, Mistral Large 4 costs about a nineteenth of GPT-6 Astra and an eighth of Claude Opus 5.5.
$1.03
blended per 1M tokens at the launch sale
Open the cost calculator- Large 4$1.03
- DeepSeek V4 Pro$1.98
- GLM-5.3$2.15
- Qwen3.8 Max$3.00
- Claude Sonnet 5.5$4.00
- GPT-6.1 Sol$4.00
- Claude Opus 5.5$8.00
- GPT-6 Astra$20.00
Prices from each provider’s pricing page and OpenRouter, checked Oct 6, 2026. DeepSeek V4 Pro at its standard (peak-hour) rate. Large 4 list price is $1.36 / $4.18; the launch sale is $0.68 / $2.09.
Use cases
What Mistral Large 4 does best
Five jobs Mistral Large 4 handles well, starting with the ones where it leads the benchmarks. Each one comes with a starter question you can send in one click.
Cybench: 93%Security work
Security is where Large 4 stands out most: 93% on Cybench, 82% on CyberGym-E2E and 50 solved tasks on the AA Cyber Index, ahead of GPT-6 Astra (33) and Claude Opus 5.5 (29). Use it to review code, read logs and explain how a bug can be fixed.
Harvey legal: #6 of 75Legal and finance documents
On Harvey’s Legal Agent Benchmark, run by Vals AI, Large 4 scores 15.8, ranking #6 of 75 models and nearly three times GPT-6 Astra’s 5.4. It also scores 54.7 on Finance Agent v2, just ahead of GPT-6 Astra.
Dense200: 42.0Images and documents
Large 4 reads text and images in one prompt through a 1.6B vision encoder. It scores 42.0 on Dense200, edging GPT-6 Astra at pointing to objects in crowded images, and 63.1 on ChartQA Pro for reading charts. Add a photo in the chat and ask it to box what you are looking for.
DeepSWE v1.1: 61.7%Coding agents
Mistral reports 61.7% on DeepSWE v1.1 and ranks Large 4 ahead of DeepSeek V4 Pro and Qwen3.8 Max on its Coding Agent Index. Function calling and structured JSON output are built in.
160+ languages160+ languages
Mistral trained Large 4 on more than 160 languages, including every official EU language, so you can draft, translate and answer support tickets without switching models.
Developers
Call Mistral Large 4 in three lines
The API uses the model name mistral-large-4 on the standard chat completions endpoint. Add reasoning_effort: "high" when you want step-by-step thinking.
- Endpoint: POST https://api.mistral.ai/v1/chat/completions
- Model: mistral-large-4 (alias mistral-large-4-0)
- Also on OpenRouter as mistralai/mistral-large-4-0
curl https://api.mistral.ai/v1/chat/completions \
-H "Authorization: Bearer $MISTRAL_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "mistral-large-4", "reasoning_effort": "high",
"messages": [{"role": "user", "content": "Hello"}]}'Timeline
Release timeline
Dec 2, 2025
Mistral Large 3
675B parameters, 41B active, Apache 2.0.
Oct 6, 2026
Large 4 public preview
API opens on Mistral Studio, OpenRouter and Vercel AI Gateway the same day.
Oct 6, 2026
50% launch sale
$0.68 input and $2.09 output per 1M tokens, down from $1.36 and $4.18.
End of Oct 2026
Open weights
Mistral says by the end of October. The Hugging Face page lists Oct 31; some press reports Oct 27.
Guides
Guides people read most
- API PricingMistral Large 4 API costs $1.36/$4.18 per 1M tokens, now 50% off at $0.68/$2.09. Estimate your bill and compare with DeepSeek V4, GPT-6 and Claude.
- BenchmarksMistral Large 4 benchmark scores: 61.7% DeepSWE, 59.9% AutomationBench, AA Index 38, Vals Index 48.05%. Mistral and independent results side by side.
- API GuideUse the Mistral Large 4 API with model ID mistral-large-4: cURL, Python and TypeScript code, reasoning_effort, image input, tools, JSON and OpenRouter.
- Run LocallyMistral Large 4 weights need about 1.05 TB at FP8 and 525 GB at 4-bit. GPU setups that fit, KV cache math and what to do before the weights ship.
- Weights & LicenseMistral Large 4 open weights are due by the end of October 2026 at mistralai/Mistral-Large-4.0-1T05-A52B. Status, release date, license and download size.
- Use It FreeYes: chat with Mistral Large 4 free on our homepage with a free account, 15 credits a day (about 15 short replies). Every other free option compared.
- vs DeepSeek V4Mistral Large 4 vs DeepSeek V4 Pro: near-identical index scores, image input vs text-only, 524K vs 1M context, and what each costs at peak and off-peak.
- vs Large 3Mistral Large 4 vs Large 3: 1.05T vs 675B parameters, built-in reasoning, 2x the context, 160+ languages, and what the higher price means for your bill.
- UpdatesDated Mistral Large 4 timeline: Oct 6, 2026 preview, launch pricing and 50% sale, OpenRouter and Vercel access, first scores, end-of-October weights.
- FAQShort, sourced answers about Mistral Large 4: parameters, context window, API price, weights date, license, image input, hardware and free access.
Plans
Mistral Large 4 Chat Pricing: Free & Pro Plans
Mistral Large 4 chat pricing: a free account with 15 credits every day, or Pro at $39.90/month or $199.90/year for long conversations and replies.
Free
$0
No card needed
- Sign up with Google or email in seconds
- 15 credits every day with a free account
- Chat history saved to your account
- Short and medium-length conversations
Pro
Best value · Save 58%$39.90$16.66/mo
Billed $199.90 once a year · 36,000 credits added right away
You save $278.90 a year vs. paying monthly
- 36,000 credits per year (about 1 credit per typical reply)
- Up to 200K tokens of context and 16K-token replies
- No daily message cap
- Interrupted replies are only charged for the text you received
- Cancel anytime in your account
- Secure checkout by Stripe
- Cancel anytime in your account
- 15 free credits a day, on every plan
FAQ
Mistral Large 4 chat FAQ
How the free chat, credits and Pro plan work.
Ask Mistral Large 4 something real
Paste a bug, a contract clause or a messy spreadsheet note. A free account gives you 15 credits every day.
Start chatting