mistral-ai

Introducing Text-Completion Codestral: Your Go-To Solution for Code Generation

Tal Peretz

— 1 min read
Introducing Text-Completion Codestral: Your Go-To Solution for Code Generation

Welcome to the world of seamless code generation with the Text-Completion Codestral/codestral-2405 model, a cutting-edge offering from Mistral AI. This model is meticulously designed to cater to a variety of code generation tasks, making it a must-have tool for developers working across different programming languages.

Model Overview

The Codestral model is optimized for code generation tasks, including fill-in-the-middle and code completion. Trained on over 80 programming languages, it boasts a plethora of capabilities:

  • Code generation
  • Code completion and suggestions
  • Code translation
  • Code summarization and explanation
  • Code refactoring, including bug fixes and test case generation

API Usage

The model can be accessed via the Mistral AI API. Here’s how you can use it:

Endpoint

To interact with the model, use the createChatCompletion or createFIMCompletion endpoints.

Request Parameters

  • Model: Specify as "codestral-2405".
  • Prompt: Provide the initial code snippet or prompt.
  • Hyperparameters: Adjust parameters like temperature, top_p, max_tokens, min_tokens, seed, and stop to control the response.
  • Streaming: Choose whether to stream the response or receive it all at once.

Example Request

{
  "model": "codestral-2405",
  "temperature": 0.7,
  "top_p": 1,
  "max_tokens": 10,
  "min_tokens": 0,
  "stream": false,
  "stop": "string",
  "random_seed": 0,
  "prompt": "def is_odd(n): \n return n % 2 == 1 \ndef test_is_odd():",
  "suffix": "return True"
}

Expected Response

The response will include details such as the completion ID, model used, usage metrics (prompt tokens, completion tokens, total tokens), and the generated code:

{
  "id": "b41e0df599f94bc1a46ea9fcdbc2aabe",
  "object": "text_completion",
  "created": 1589478378,
  "model": "codestral-latest",
  "choices": [
    {
      "text": "\n assert is_odd(1)\n assert",
      "index": 0,
      "logprobs": null,
      "finish_reason": "length"
    }
  ],
  "usage": {
    "prompt_tokens": 5,
    "completion_tokens": 7,
    "total_tokens": 12
  }
}

Additional Notes

Access to the model requires an API key. Integration is seamless across various platforms, including Vertex AI. For more detailed documentation, refer to the official Mistral AI API documentation.

Read more