On this page
The short version
What is Le Chonk?
The nickname for Mistral Large 4, a large mixture-of-experts model. You can try its API preview today.
How much does it cost?
The model docs display $0.68 for input and $2.09 for output per million tokens. Higher rates are listed too.
Can I download the weights?
Not from a release we have verified. A public weight release has not been verified here.
Will it run on my GPU?
There is no tested setup here yet. Active parameters tell you about computation, not total weight storage.
Model specifications
Start with the model version. Le Chonk is a public preview, so a result measured today may not describe a later release.
| Specification | What we know |
|---|---|
| Architecture | Mixture of experts (MoE) |
| Total parameters | About 1T in the release; 1.05T in model documentation |
| Active parameters | 49B in the release; 52B in model documentation. Difference not explained. |
| Context window | 1M tokens, as listed in the model documentation |
| Input / output | Multimodal input; text output. Check the endpoint for supported formats. |
| API status | Public preview · model ID listed as mistral-large-4 |
| Downloadable weights | Pending verified release |
| License | Not verified. Check the license attached to the actual weight release. |
The 49B / 52B difference comes from two official pages. We keep both figures visible instead of guessing why they differ.
What would your API bill look like?
Put in your usual request size. The calculator applies the same token counts to each provider, so you can see the price difference. Actual token counts can differ between models.
10,000 requests per month
| Model / provider | Input / output¹ | Per request | Monthly | vs Le Chonk² |
|---|---|---|---|---|
| Le ChonkMistral AI ↗ | $0.68 / $2.09 | $0.00345 | $34.50 | Baseline |
| Qwen3.8-MaxQwenCloud ↗ | $2.00 / $6.00 | $0.01000 | $100.00 | +$65.50 |
| DeepSeek Pro · off-peakDeepSeek direct API ↗ | $0.66 / $1.98 | $0.00330 | $33.00 | −$1.50 |
| DeepSeek Pro · peakDeepSeek direct API ↗ | $1.32 / $3.96 | $0.00660 | $66.00 | +$31.50 |
| DeepSeek Flash · off-peakDeepSeek direct API ↗ | $0.15 / $0.60 | $0.00090 | $9.00 | −$25.50 |
| DeepSeek Flash · peakDeepSeek direct API ↗ | $0.30 / $1.20 | $0.00180 | $18.00 | −$16.50 |
¹ USD per million uncached tokens. ² Monthly difference; a minus sign means cheaper. No caching, batch discounts, tool fees, taxes or tokenization differences. DeepSeek peak and off-peak rates are shown separately. Mistral also lists higher $1.36 / $4.18 rates; no discount end date confirmed.
How the calculation works
Monthly cost = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000. The default example is 2,000 input tokens, 1,000 output tokens and 10,000 requests: $0.00345 per request, or $34.50 a month for Le Chonk.
Benchmarks, with their sources
These scores are reported in Mistral’s announcement and attributed to Artificial Analysis. We have not reproduced the tests or directly checked the evaluator’s records. Read them as reported results, not our own measurements.
| Benchmark | Result | Evidence |
|---|---|---|
| DeepSWEv1.1 | 61.7% | Vendor-reportedMistral → Artificial Analysis ↗ |
| SWE-Atlas-QnANot specified in release | 59.4% | Vendor-reportedMistral → Artificial Analysis ↗ |
| Terminal-Bench4.0 | 28.3% | Vendor-reportedMistral → Artificial Analysis ↗ |
| Coding Agent IndexNot specified in release | 49.8% | Vendor-reportedMistral → Artificial Analysis ↗ |
Le Chonk vs other models
A cheaper token does not always mean a cheaper finished task. Retries, long answers and failed tool calls can change the bill. The comparison pages keep price and performance separate.
Can you run Le Chonk locally?
There is no verified local setup in this guide yet. Calling Mistral’s API from a laptop still runs the model on Mistral’s servers. Local inference means downloading and loading the weights yourself.
49B active does not mean 49B weights to store. Using the documented 1.05T total, even raw 4-bit weights work out to about 525 GB. That excludes memory needed to run the model.
A few things worth clearing up
Is Le Chonk the same as Mistral Large 4?
Yes. Le Chonk is the nickname used in Mistral’s release announcement. This guide refers to that public preview.
Is it open source?
Mistral describes an open-weight model and says weights will be released. We have not checked a final weight license. Open weights and open source are not interchangeable terms.
Why are there two API prices?
The model documentation displays $0.68 / $2.09 alongside $1.36 / $4.18. This calculator uses the displayed lower rates. We have not confirmed when those rates might end; check your provider before budgeting.
Can a 24 GB GPU run it?
We have no verified result for that setup. Offloading weights to system memory may change what is possible, but also affects speed. A memory estimate is not proof that a configuration works.
Which model should I choose?
Try the same small set of real tasks on each model. Track accepted outputs, retries, time and total spend. A benchmark score alone cannot tell you which model will work best in your app.
Sources
- Mistral Large 4 release announcement
API preview, planned weight release, 49B active parameters and reported benchmark results.
- Mistral Large model documentation
1.05T total, 52B active, 1M context. Displays both $0.68 / $2.09 and higher $1.36 / $4.18 prices. No discount end date confirmed.
- Qwen3.8-Max model and pricing
QwenCloud endpoint qwen3.8-max. USD uncached input/output rates; not a quote for every region or reseller.
- DeepSeek models and pricing
Direct API rates. Separate Pro 0813 and V4.1 Flash. Peak hours: weekdays 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays.