Codestral
At a glance
- Price per 1M tokens (input)
- $0.30
- Price per 1M tokens (output)
- $0.90
- Context
- 128,000 tokens
- Free access
- yes (see below)
Refreshed daily; data verified September 29, 2026. Published August 12, 2026, last price change August 30, 2026.
Codestral is a code-specialized language model from Mistral AI, designed for developers and engineering teams who need to generate, review, and refactor source code within their existing workflows. Its primary strength lies in tool calling (function calling), which allows the model to invoke external APIs and development tools directly, making it suitable for tasks like automated code completion, test generation, and CI/CD pipeline assistance. Unlike general-purpose models, Mistral's approach focuses on deep integration with programming contexts, prioritizing accuracy and efficiency in code-centric operations. This model is for teams that want to embed AI assistance into their development environment rather than use a one-off chat interface.
Specifications & pricing
| Input (per 1M tokens) | $0.30 |
|---|---|
| Output (per 1M tokens) | $0.90 |
| Cache read (per 1M tokens) | $0.03 |
| Context window | 128,000 tokens |
| Max output | 128,000 tokens |
| Capabilities | tool calling |
LiteLLM community dataset (MIT), verified September 29, 2026. Official Mistral AI pricing.
What Codestral would cost on your workload — run it through the cost calculator →
Price history
| Date | Input / 1M | Output / 1M |
|---|---|---|
| August 9, 2026 (first record) | $1.00 | $3.00 |
| August 22, 2026 | $0.30 | $0.90 |
| August 30, 2026 | $0.30 | $0.90 |
Where to try Codestral for free
- Mistral AI offers a free chat — Le Chat (free plan). A vendor's free chat may run a different model from the same family — the exact model is not guaranteed.
Frequently asked questions
What is Codestral best used for in a business setting?+
Codestral is best for automating and accelerating software development tasks, such as generating boilerplate code, writing unit tests, explaining existing code, and assisting with code reviews. Its tool calling capability means it can interact with your development tools and services, enabling more complex workflows like triggering builds or fetching issue data.
How do I get started with Codestral?+
You can access Codestral through Mistral AI's platform or via API. For tool calling, you'll need to set up your environment to allow the model to make function calls, typically by defining functions in your request and handling the model's responses to execute those functions. The documentation provides examples for major programming languages.
How does Codestral differ from other Mistral models?+
Codestral is specialized for code, whereas Mistral's other models are more general-purpose or tuned for chat. Codestral is optimized for code generation and understanding, with a focus on tool calling to support developer workflows. Other models may handle broader language tasks but lack the same depth in code-specific patterns and tool integration.
What are the limitations of Codestral?+
Codestral may not perform as well on non-code tasks, such as general knowledge or creative writing, compared to general-purpose models. Its tool calling requires careful setup and error handling, and it may occasionally produce incorrect or insecure code, so human review is still essential. Additionally, it may struggle with very large codebases or highly niche programming languages.
Compare with others
Org chart: how to move your company onto AI
A practical map: which company roles and processes AI agents can take over, where to start, and in what order to roll it out.