Codestral 2508

Codestral 2508 serves as a specialized language model within the Mistral ecosystem, purpose-built to address the rigorous demands of software development and programming workflows. Its architecture is specifically optimized for low-latency, high-frequency tasks, making it a robust tool for developers who require rapid code generation and precise completion. By focusing on the nuances of syntax and logic, the model is designed to handle complex coding environments, providing a reliable foundation for tasks that require deep technical understanding and structural accuracy.

The model demonstrates significant practical utility in specialized programming functions, including fill-in-the-middle capabilities, automated code correction, and the generation of comprehensive test suites. Its design lineage emphasizes efficiency and performance, allowing it to process large codebases and complete project files with ease. As a flagship entry in its family, Codestral 2508 is positioned to support modern development pipelines, offering a streamlined approach to technical challenges where precision and speed are paramount for maintaining high-quality software output.

OpenRoutermistralai/codestral-2508codestral

Open provider page Provider docs

Quick Info

Provider: OpenRouter
Model key: mistralai/codestral-2508
Release date: Aug 1, 2025
Last updated: Aug 1, 2025
Knowledge cutoff: 2025-03-31
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.30
Output token cost: $0.90

Limits

Output tokens: 204,800 tokens
Context window: 256,000 tokens

Transparent token rates

Compare Codestral 2508 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Codestral 2508

OpenRouter

Coverage

AICoder reports (dated 2026-08-30) that Mistral AI released Codestral 25.08 as its latest flagship open coding model, describing a native 256k token context window designed to ingest entire repository trees for symbol-level resolution, along with an enhanced Fill-in-the-Middle (FIM) optimization that the article says d According to the same report, Codestral 25.08 weights are published on Hugging Face, with hosted API endpoints available and an instant-setup path through Ollama, giving developers multiple deployment options. The article frames this as Mistral AI's flagship open coding update, tying repo-level completion, FIM latency,

OpenRouter

CoverageBenchmark

Kendr's evidence profile for mistralai/codestral-2508 (snapshot updated 2026-09-06) lists a 256,000-token context window, text plus file input modalities, and declared capabilities including file input, prompt caching, structured output, text generation, and tool calling, with Mistral AI listed as the model developer. The same profile records reference pricing of $0.30 per 1M input tokens and $0.03 per 1M cached input tokens from the source snapshot, and notes a knowledge cutoff of 2025-03-31 with the entry cataloged on August 1, 2025. The page is presented as a research/comparison profile rather than a callable route, so the facts