Codestral Embed
Generally availableProprietary8K context
Summary
Codestral Embed is a proprietary language model from Mistral AI, released on 28 May 2025. It accepts text and generates text, with a 8K-token context window and up to 3K output tokens per response.
Embedding model for code search, retrieval, and code-aware RAG pipelines
Last verified 21 Oct 2026. Every figure on this page links to its source.
Specifications
| Developer | Mistral AI |
|---|---|
| Release date | 28 May 2025 [models.dev] |
| Status | Generally available [models.dev] |
| Weights | Proprietary (API only) [models.dev] |
| Context window | 8K (8,192 tokens) [models.dev] |
| Max output | 3K tokens [models.dev] |
| Input | Text [models.dev] |
| Output | Text [models.dev] |
| Reasoning mode | No [models.dev] |
| Tool calling | No [models.dev] |
Codestral Embed API pricing by provider
USD per million tokens. Sorted by input price.
| Provider | Input | Output | Cached input | Context | Source |
|---|---|---|---|---|---|
| Vercel AI Gateway mistral/codestral-embed | – | – | – | 8K | models.dev, 21 Oct 2026 |
Frequently asked questions
What is the context window of Codestral Embed?
Codestral Embed has a context window of 8,192 tokens (8K) and can generate up to 3,072 tokens in a single response, according to models.dev.
Is Codestral Embed open source?
No. Codestral Embed is a proprietary model; Mistral AI has not released its weights. It is available through APIs including Vercel AI Gateway.
When was Codestral Embed released?
Mistral AI released Codestral Embed on 28 May 2025, according to models.dev.
Change history
- Max output changed from 2K to 3K
- Added to the directory
- Now listed on Vercel AI Gateway