Legacy model · Mistral AI
Codestral 22B benchmarks, pricing and specs
Codestral 22B is a Mistral AI model released on May 29, 2024 that takes text input and produces text. It is now listed as a legacy model. Its price is not published and accepts up to 32,000 tokens of context.
- Context window
- 32K#68 of 68 current models
Data verified August 6, 2026 · Dataset 2026-11-06
Specs
Key facts
- Provider
- Mistral AI
- Model family
- Codestral
- Release date
- May 29, 2024
- Status
- Legacy
- Access
- API and open weights
- API model id
- Not reported
- Context window
- 32,000 tokens
- Max output
- Not reported
- Input
- text
- Output
- text
- Task families
- text
- Declared capabilities
- reasoning
Charts
Against every current text model
Where Codestral 22B lands among the current models of its task, on every independent benchmark, on price and on context.
Where it stands
Codestral 22B is no longer a current model, so it is not ranked; each bar is the share of today's text models its value beats. Each bar is the share of current models it beats.
Price and context
Context window
Not rankedno longer current
32K tokens · longer than 0%
Benchmarks
Independent benchmark results
Scores come from independent evaluators that ran the model themselves, pinned to a dated snapshot. Rank is among current models in the same snapshot. Mistral AI's own launch figures are listed under strengths with the publisher named, because they are not measured the same way.
No independent benchmark has published a result for Codestral 22B yet. It is still compared below on price, context and capabilities.
Fit
Strengths, cautions and best uses
- Strengths
- Codestral 22B records both API and open-weight access for text inputs and a published 32,000-token context window
- Cautions
- This record is marked legacy; confirm current availability and migration requirements before adoption
- Best for
- Code-generation trials with API or open-weight deployment
From benchmark to production
Building on Codestral 22B?
Public scores rarely hold on your own data. We test it against the alternatives on your real tasks, then build the agent or pipeline around the model that wins.
AI Model Cost AuditLaunch price $100, regular price $150What the audit includes
Alternatives
Alternatives to Codestral 22B
The closest current text models by price. The cheapest model that scores as well or better is listed first; failing that, the cheapest one within 3% of the benchmark's scale.
On score and price
Codestral 22B beside the 4 alternatives listed below, on Arena Text.
Arena Text
Measured October 28, 2026, higher is better
- Codestral 22B
- Not scored
- DeepSeek-R1
- 1,398
- Qwen2.5-VL 72B Instruct
- Not scored
- Kimi-VL-A3B-Instruct
- Not scored
- GPT-4.1
- 1,415
DeepSeek-R1
- 128K context window vs 32K
- From DeepSeek
Qwen2.5-VL 72B Instruct
- 128K context window vs 32K
- From Alibaba Qwen
Kimi-VL-A3B-Instruct
- 128K context window vs 32K
- From Moonshot Kimi
GPT-4.1
- 1.05M context window vs 32K
- From OpenAI
FAQ
Codestral 22B FAQ
What is the context window of Codestral 22B?
Codestral 22B accepts up to 32,000 tokens of context. That ranks #68 of 68 current text models by context length.
Is Codestral 22B open source or API only?
Codestral 22B is available both through Mistral AI's API and as open weights you can run yourself.
What are the best alternatives to Codestral 22B?
The closest current alternatives by price and measured performance are DeepSeek-R1 (128K context window vs 32K), Qwen2.5-VL 72B Instruct (128K context window vs 32K), Kimi-VL-A3B-Instruct (128K context window vs 32K) and GPT-4.1 (1.05M context window vs 32K).
Sources
Sources
- Codestral · Mistral AI · official · verified August 6, 2026
- Mistral AI pricing · Mistral AI · official · verified August 6, 2026
All 255 sources behind the benchmarks, with their versions, licenses and notes, are in the source catalog.
Cite this page
Van Data Team. "Codestral 22B Benchmarks, Pricing and Specs." AI Model Benchmark Dataset 2026-11-06. https://vandatateam.com/ai-model-benchmarks/codestral-22b
Reuse the data: JSON · CSV · or explore every model in the AI model benchmark comparison.
Production evaluation
Deciding whether Codestral 22B fits your workload?
Public benchmarks narrow the shortlist. We test the finalists on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.
- Evaluation cases and acceptance thresholds tied to real work
- Model routing, fallbacks and cost controls
