Mistral AI · Released May 29, 2024Codestral
Mistral code model for completions, refactors, and developer IDE workflows.
- Capability
- —No ECI score yet
- Input
- $0.30per 1M tokens
- Output
- $0.90per 1M tokens
- Context
- 256K4K max output
What it costs
- Input
- $0.30per million tokens
- Output
- $0.90per million tokens
- Cached input
- $0.03per million tokens
- Blended (3:1)
- $0.45cheaper than 63% of priced models
Typical monthly bills
| Monthly usage | Estimated cost |
|---|---|
| Side project2M input + 0.5M output tokens | $1.05 |
| Team assistant25M input + 5M output tokens | $12.00 |
| Production app250M input + 50M output tokens | $120.00 |
Official Mistral API price, as listed on models.dev.
Independent benchmarks
Epoch AI has not published a Capabilities Index score for Codestral yet.
The details
- Lab
- Mistral AI
- Released
- May 29, 2024
- Knowledge cutoff
- Oct 2024
- Context window
- 256,000 tokens
- Max output
- 4,096 tokens
- Inputs
- Text
- Output
- Text
- Reasoning
- No
- Tool calling
- Yes
- Structured output
- No
- Weights
- Open weights
- API model ID
codestral-lateston Mistral- Availability
- 3 API providerslisted on models.dev
What Mistral AI claims
Published by the lab at launch. Settings vary, so compare these only with care.
| Benchmark | Score | Setting | Source |
|---|---|---|---|
| Aider Polyglot | 11.1 percent correct | — | Source |
More from Mistral AI
About Codestral
How much does Codestral cost?
Codestral costs $0.30 input / $0.90 output per million tokens (official Mistral API price). Cached input is $0.03 per million tokens. At a 3:1 input-to-output mix that is $0.45 per million tokens, cheaper than 63% of the 360 priced models we track.
What is the context window of Codestral?
Codestral accepts up to 256,000 tokens per request and can write up to 4,096 tokens in one response.
How good is Codestral?
Epoch AI has not published a Capabilities Index score for Codestral yet.
Is Codestral open source?
Yes. Mistral AI publishes the weights, so it can be downloaded and self-hosted.
What inputs does Codestral support?
Codestral accepts text and replies in text. It is not a dedicated reasoning model, supports tool calling.
When was Codestral released?
Mistral AI released Codestral on May 29, 2024. Its training data runs to Oct 2024.
What are the best alternatives to Codestral?
The closest current models from other labs on capability, price and release date are Qwen2.5-VL 7B Instruct (Alibaba (Qwen), $0.35 / $1.05), Command R (Cohere, $0.15 / $0.60), Llama-3.2-11B-Vision-Instruct (Meta, $0.197 / $0.51) and Llama 3.1 Nemotron 70B Instruct (NVIDIA, $0.478 / $0.504).