Mistral: Codestral 2508

Text input Text output
Author's Description

Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correction and test generation. [Blog Post](https://mistral.ai/news/codestral-25-08)

Key Specifications
Cost
$$$
Context
256K
Parameters
2.5B (Rumoured)
Released
Aug 01, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Tools Frequency Penalty Temperature Tool Choice Seed Response Format Max Tokens Structured Outputs Presence Penalty Top P Stop
Features

This model supports the following features:

Structured Outputs Tools Response Format
Performance Summary

Mistral's Codestral 2508, a coding-specialized language model, demonstrates competitive response times, performing among the faster models with a 50th percentile speed ranking across seven benchmarks. It also offers cost-effective solutions, ranking in the 65th percentile for price. The model exhibits a mixed performance across various categories. Its primary strength lies in its exceptional cost efficiency in Mathematics, achieving a top 3 ranking for cost, despite a very low accuracy (3.0%). It also shows solid performance in Instruction Following (61st percentile accuracy) and respectable cost-effectiveness in Ethics, Email Classification, Instruction Following, Reasoning, and Coding. However, Codestral 2508 struggles significantly with General Knowledge (17th percentile accuracy) and Mathematics accuracy (9th percentile). Its performance in Ethics (34th percentile accuracy) and Coding (23rd percentile accuracy) is below average, which is notable given its specialization. The model's speed in Reasoning is exceptionally high, ranking in the 98th percentile for duration. Overall, Codestral 2508 appears to be optimized for low-latency, high-frequency tasks, as indicated by its description, with a clear trade-off in general knowledge and mathematical accuracy for cost and speed in specific areas.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.3
Completion $0.9

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Mistral
Mistral | mistralai/codestral-2508 256K $0.3 / 1M tokens $0.9 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by mistralai