Codestral 22B
Codestral 22B is Mistral AI's first code-specialist model, a 22-billion-parameter dense transformer released in May 2024. It handles 80-plus programming languages — including Python, JavaScript, Rust, and Go — and ships with 32K context, which covers most file-level and multi-file refactor tasks without chunking. On HumanEval it outperformed same-size generalist models and competes directly with DeepSeek Coder V2 Lite and Qwen 2.5 Coder 7B for the under-50B coding-model slot. The catch: Codestral ships under the Mistral Research License, which blocks commercial use without a separate agreement from Mistral. If your workload is production-facing, budget for that licensing conversation before committing to the stack.
No pricing data available for this model yet.
Not enough history to display chart. View full history →
No benchmark data available.
Questions developers ask
How much does it cost to run Codestral 22B for 100M tokens?▾
Pricing varies by provider. Use the workload calculator to get an accurate estimate for your specific usage pattern.
What is the cheapest provider for Codestral 22B?▾
No pricing data is available yet. Check back after the next scrape run.
What context window does Codestral 22B support?▾
Codestral 22B supports a context window of 32,768 tokens. Individual providers may cap this lower.
What's the cheapest way to run Codestral 22B?▾
Compare provider prices in the table above and consider using prompt caching if supported.
Is there a free tier for Codestral 22B?▾
Free tiers vary by provider and change frequently. Check each provider's current pricing page for trial credits or free-tier limits. The prices shown on this page reflect paid API access.
How much does Codestral 22B cost for 1M tokens?▾
Pricing data for Codestral 22B is not yet available. Check back after the next data refresh.
Keep exploring
No hosted pricing scraped for this model yet. Methodology · Raw data