Mistral AI, proprietary
Codestral
Codestral by Mistral AI ranks 290th of 354 ranked models on the Noometry Index as of October 2026, with a score of 30.6. Its strongest category is reasoning, where it ranks 251st. API pricing starts at $0.30 per million input tokens and $0.90 per million output tokens, with a 256K-token context window.
Last verified
Specifications
- Noometry rank
- #290 of 354
- Index score
- 30.6
- Evidence
- Reported 7 results
- Provider
Mistral AI- Released
- May 29, 2024
- Weights
- Proprietary
- Reasoning
- No
- Context window
- 256K
- Max output
- 8K
- Input price
- $0.30 / M
- Output price
- $0.90 / M
- Blended price
- $0.45 / M
- Output speed
- 271 tokens/s Kagi
- Value
- #85 of 219
- Knowledge cutoff
- October 2024
- Input
- text
Category scores
Each category score combines every public result we have in that category.
- Coding 27.3
- Reasoning 19.8
Strengths and weaknesses
Categories where Codestral places highest and lowest among the models ranked in each, with its score against that category's median.
Closest competitors
The models ranked just above and below Codestral. When scores are this close, price and speed are often the better way to choose.
| Model | Rank | Score | Blended $/M | Speed | |
|---|---|---|---|---|---|
| Granite 3.0 2b Instruct | #286 | 30.8 | — | — | Compare |
| Codellama 34b Instruct | #287 | 30.8 | — | — | Compare |
| Llama 3.1-405B | #288 | 30.7 | — | 78 | Compare |
| Yi-1.5-34B | #289 | 30.6 | — | — | Compare |
| Llama-3.3-70B-Instruct | #291 | 30.6 | $0.16 | — | Compare |
| GPT-4 Turbo | #292 | 30.5 | $15 | — | Compare |
| Qwen1.5-32B | #293 | 30.5 | — | — | Compare |
| Amazon Nova Micro | #294 | 30.4 | $0.0613 | — | Compare |
Sponsored placements are available on pages like this one. Advertise on Noometry
Benchmark results
Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.
Coding
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Aider Polyglot | 11.1% | #41 of 44, top 94% | Epoch AI | ||
| BigCodeBench Instruct | 41.8% | #30 of 64, top 47% | BigCodeBench | 2024-05-23 | |
| BigCodeBench Complete | 52.5% | #26 of 66, top 40% | BigCodeBench | 2024-05-23 | |
| ALE-Bench | 137.78 | #105 of 105, top 100% | Epoch AI | ||
| HumanEval+ | 73.8% | #16 of 45, top 36% | EvalPlus | ||
| MBPP+ | 61.9% | #22 of 38, top 58% | EvalPlus |
Reasoning
| Benchmark | Score | Position | Setting | Source | Date |
|---|---|---|---|---|---|
| Kagi LLM Benchmark | 32.5% | #91 of 99, top 92% | Kagi LLM Benchmark |
API pricing by provider
| Route | Input $/M | Output $/M | Cached input $/M | Checked |
|---|---|---|---|---|
| azure | $0.30 | $0.90 | — | 2026-10-10 |
| mistral | $0.30 | $0.90 | $0.03 | 2026-10-10 |
| openrouter | $0.30 | $0.90 | $0.03 | 2026-10-10 |
Compare Codestral
- Codestral vs Yi-1.5-34B
- Codestral vs Llama-3.3-70B-Instruct
- Codestral vs Llama 3.1-405B
- Codestral vs GPT-4 Turbo
- Codestral vs Codellama 34b Instruct
- Codestral vs Qwen1.5-32B
- Codestral vs GPT-6 Astra
- Codestral vs Claude Fable 5.1
- Codestral vs Gemini 3.8 Flash
- Codestral vs Kimi K3
- Codestral vs Grok 4.6
- Codestral vs Qwen3.8 Max
- Codestral vs GLM-5.3
- Codestral vs Muse Spark 1.3
Other Mistral AI models
- Mistral Large 443.1
- Mistral Medium 3.540.2
- Mistral Large 339.1
- Mistral Medium36.3
- Magistral Medium35.2
- Devstral Small 250534.3
- Mistral Small33.4
- Pixtral Large32.2
Frequently asked questions
How good is Codestral?
Codestral by Mistral AI ranks 290th of 354 ranked models on the Noometry Index as of October 2026, with a score of 30.6. Its strongest category is reasoning, where it ranks 251st. API pricing starts at $0.30 per million input tokens and $0.90 per million output tokens, with a 256K-token context window.
How much does Codestral cost?
Codestral costs $0.30 per million input tokens and $0.90 per million output tokens on Mistral AI's own API, with cached input at $0.03.
What is Codestral's context window?
Codestral accepts up to 256K tokens of input and can write up to 8K tokens in one response.
Is Codestral open source?
No. Codestral is proprietary and available only through Mistral AI's API and partner platforms.
How fast is Codestral?
Codestral generated about 271 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.
What are Codestral's strengths and weaknesses?
Relative to other ranked models, Codestral places best in reasoning and lowest in coding.
What is Codestral best at?
Its best category is reasoning, where it ranks 251st on Noometry.