Log in
← All models

Codestral

Budget

Mistral's fast code-completion model — cheap and quick for inline assistance.

Mistral
Maker
72/100
Overall
70
Value score
$0.48
Blended /1M
256K
Context
Proprietary
Weights
Pricing: $0.30 in · $0.90 out / 1M tokens · synced from OpenRouter · updated 2026-06-22

Scorecard

Output Quality6.8/10 · 67% community
Agentic Ability6.5/10 · 65% community
Speed9.5/10 · 79% community
Value for $8.9/10 · 77% community
Reliability8.8/10 · 76% community

Codestral (2508) is Mistral's fast, low-latency code model with a 256K context, aimed at autocomplete and quick edits rather than heavy agentic planning. Cheap and snappy where speed beats depth.

Strengths
  • Very fast for code
  • Cheap
  • 256K context
Trade-offs
  • Not for deep agentic planning
  • Quality below the larger coders

Agents that run Codestral well

Fast completions and lightweight code edits.

Best model for ContinueBest model for Kilo CodeBest model for ClineBest model for Aider

Codestral FAQ

Is Codestral good for AI agents?
Fast completions and lightweight code edits. On our editorial scoring it rates 6.5/10 for agentic ability and 6.8/10 for output quality. Very fast for code.
How much does Codestral cost?
Via OpenRouter, Codestral is priced at $0.30 in · $0.90 out / 1M tokens — a blended rate of about $0.48 per 1M tokens for typical input-heavy agent use.
What is Codestral's context window?
Codestral has a 256K-token context window. Mistral is the maker.
Can I self-host Codestral?
No — Codestral is a proprietary, hosted-only model from Mistral. You access it through an API (e.g. OpenRouter) rather than running the weights yourself.
What are the downsides of Codestral?
The main trade-offs: not for deep agentic planning; quality below the larger coders. It's strongest for fast completions and lightweight code edits.

Codestral compared

Codestral vs Haiku 4.5Codestral vs DeepSeek V4 FlashCodestral vs GPT-5.4 MiniCodestral vs Qwen3 235BCodestral vs Flash LiteCodestral vs Llama 4 MaverickCodestral vs Devstral 2Codestral vs Mistral Large 3

The model is half the story — the agent is the other half

The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.

Compare AI agents →