Opening soon. One email when chat opens, nothing else. · Independent site, not affiliated with Mistral AI. AI can be wrong.
Mistral Le Chonk at a glance
Le Chonk is the nickname Mistral AI gave to Mistral Large 4. These are the facts reported at launch, with the source of each.
| Official name | Mistral Large 4 (nickname: Le Chonk) |
|---|---|
| Announced | 6 October 2026 by Mistral AI |
| Size | About 1 trillion parameters, 49 billion active per token (mixture of experts) |
| Input | Text and images (natively multimodal) |
| Languages | 160+, including every official EU language |
| API model ID | mistral-large-4 (public preview) |
| Preview API price | $0.68 per million input tokens, $2.09 per million output tokens |
| List API price | $1.36 per million input tokens, $4.18 per million output tokens |
| Open weights | Promised for late October 2026; reports give 27 or 31 October |
Sources: Mistral AI announcement and documentation, TechCrunch, The Register, 6–8 October 2026. Preview terms can change.
What is Mistral Le Chonk?
Mistral Le Chonk is Mistral AI’s newest flagship language model. Its official name is Mistral Large 4, and “Le Chonk” is the playful nickname the company used in its launch post. It is a mixture-of-experts model: about one trillion parameters in total, but only 49 billion of them are used for each token, which keeps it much cheaper to run than a dense model of the same size.
Mistral says it is the best open-weights model from the United States or Europe on aggregated benchmarks, and that it is strong on cyber defence, manufacturing and finance work. It reads images as well as text. Mistral trained it for about two months on 4,000 Nvidia Grace Blackwell GPUs in its European data centres. Independent testers have already questioned some of the headline claims, for example the advertised one-million-token context window, so treat launch numbers as claims until more testing is published.
The weights are not public yet. Mistral says it will release them after safety testing, around the end of October 2026. Until then the model is available through Mistral’s own API in a public preview, and that is what ChonkChat uses.
How to use Mistral Le Chonk on ChonkChat
Chat is opening soon. When it does, you can be chatting with Mistral Le Chonk in under a minute.
- 1
Open the chat box
Go to the top of this page. The chat box is the first thing on the screen and works on phones and desktops.
- 2
Sign in with Google
Press the sign-in button. One click creates your account and adds 100 free credits, about 10 replies. We only receive your name and email address.
- 3
Ask your question
Type a question, paste text to summarise or ask for code. Longer, clearer prompts give better answers from Mistral Le Chonk.
- 4
Watch your credits
You pay for the tokens each reply uses: 4 credits per thousand input tokens and 12 per thousand output tokens. A typical short answer costs about 10 credits, and your balance is shown under the box.
- 5
Top up when you need more
Open the credits page and buy a pack. Credits are one-off purchases, not a subscription, and they do not expire.
What Mistral Le Chonk is good for
Based on what Mistral and independent testers have published so far, these are the tasks where the model is worth trying.
Long, careful writing
Drafting, rewriting and summarising documents, reports and emails in more than 160 languages.
Code help
Explaining code, writing functions and reviewing small changes. Mistral reports strong results on real-world coding benchmarks.
Reading finance and legal text
Independent write-ups describe credible results on finance and legal-agent tests, though not on every benchmark.
Everyday questions in European languages
French, German, Spanish, Italian and the other official EU languages were a stated training focus.
Credit packs for Mistral Le Chonk
Pay per token: $4 per million input tokens and $12 per million output tokens. $1 buys 1,000 credits, and each reply deducts what it actually used, so a short question costs less than a long one.
Free
Try Mistral Le Chonk before you pay.
- 100 credits on sign-up
- About 10 replies
- No card needed
$0
Small
Try it properly
- 5,000 credits
- 1.25M input tokens
- or 0.42M output tokens
- Credits never expire
$5once
Medium
For regular use
- 16,500 credits (+10% bonus)
- 4.13M input tokens
- or 1.38M output tokens
- Credits never expire
$15once
Large
For heavy users
- 36,000 credits (+20% bonus)
- 9M input tokens
- or 3M output tokens
- Credits never expire
$30once
Pay as you go
Buy a pack when you need it. Nothing renews.
- Input tokens: $4 per million. Output tokens: $12 per million
- You pay for the tokens each reply actually uses
- One-off payment, no subscription
- Same Mistral Le Chonk model on every pack
Prices in US dollars. Payments are handled by Stripe. We do not store card details.
Mistral Le Chonk: where to use it
There are three ways to reach the model today. This is how they differ.
| ChonkChat | Mistral Le Chat | Mistral API | |
|---|---|---|---|
| Who runs it | Independent site | Mistral AI | Mistral AI |
| Sign-up | Google, one click | Mistral account | Developer account and key |
| Cost | 100 free credits, then pay per token | Check Mistral’s plans | Pay per token |
| Best for | Trying the model quickly | Mistral’s full assistant | Building your own app |
We do not know which Le Chat plans include Mistral Large 4. Check Mistral’s own pricing page before you choose.
Mistral Le Chonk questions
Short answers to what people ask most.