Codestral 2508
Codestral 2508 is a specialized coding model developed by Mistral. It is designed for low-latency programming tasks, including code correction, test generation, and fill-in-the-middle operations.
Codestral 2508 specs and pricing
- Provider: Mistral
- Released: August 1, 2025
- Context window: 256K tokens
- Max output: 205K tokens
- Input: text, files
- Output: text
- Price: $0.30 input / $0.90 output per 1M tokens
- Tool calling: yes
What Codestral 2508 is good at
- Generating unit tests for existing codebases
- Correcting syntax and logic errors in source files
- Completing code snippets using fill-in-the-middle functionality
To-dos to try with Codestral 2508
- Analyze this script and generate a suite of unit tests for the primary functions.
- Review this code file and identify any syntax errors or potential bugs.
- Insert the missing logic into this function to complete the fill-in-the-middle task.
FAQ
- What is the primary purpose of Codestral 2508? It is a language model specifically optimized for coding tasks, such as generating tests and correcting code.
- Does this model support tool calling? Yes, Codestral 2508 includes tool calling capabilities.
- How much does Codestral 2508 cost? Codestral 2508 costs $0.30 per million input tokens and $0.90 per million output tokens through OpenRouter. In todo.is a typical agent step costs about 0.97 credits ($1 = 100 credits).
- What is the context window of Codestral 2508? Codestral 2508 can read up to 256K tokens at once and write up to 205K tokens in one reply.
- Can I use Codestral 2508 without coding? Yes. In todo.is you pick Codestral 2508 as your agent's model and write to-dos in plain words, on the web, the Android app, WhatsApp or Telegram. No API key needed: credits are included, or you can bring your own key.
JavaScript is required to use the todo.is app.