Model comparison
Codestral vs Mistral 7B
Codestral is the stronger model overall, scoring 30.1 to 22.9 on the Noometry Index. Mistral 7B costs 1.8× less per token, which makes it the better buy when Codestral's lead doesn't matter for your workload.
Last verified . 4 shared benchmarks.
Mistral AISummary
- They share 4 benchmarks with published results for both. Codestral scores higher in 2 categories and Mistral 7B in 0 categories; one gap is clear of the uncertainty.
- The widest gap is in reasoning, where Codestral leads 19.1 to 12.6.
- The biggest single-benchmark swing is BigCodeBench Complete: 52.5% for Codestral and 27.3% for Mistral 7B.
- Mistral 7B is cheaper at $0.25 / $0.25 per million input/output tokens, against $0.30 / $0.90 for Codestral.
- Codestral accepts more context: 256K tokens versus 8K.
Side by side
| Codestral | Mistral 7B | |
|---|---|---|
| Provider | Mistral AI | Mistral AI |
| Noometry Index | 30.1 | 22.9 |
| Released | 2024-05-29 | 2023-09-27 |
| Weights | Open | Open |
| Context window | 256K | 8K |
| Max output | 4K | 8K |
| Input $ / M tokens | $0.30 | $0.25 |
| Output $ / M tokens | $0.90 | $0.25 |
| Results tracked | 7 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Codestral: 26.9 (#316), Mistral 7B: 25.9 (#321)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| BigCodeBench Instruct | 41.8% | 19.5% |
| BigCodeBench Complete | 52.5% | 27.3% |
| HumanEval+ | 73.8% | 36% |
| MBPP+ | 61.9% | 42.1% |
| Aider Polyglot | 11.1% | — |
| LMArena Coding | — | 1082 |
| ALE-Bench | 137.78 | — |
Reasoning Codestral leads
Codestral: 19.1 (#244), Mistral 7B: 12.6 (#331)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | — |
| Chess Puzzles | — | 0% |
| LMArena Hard Prompts | — | 1067 |
| DTBench | — | 42.5% |
| Adversarial NLI | — | 47.1% |
| BIG-Bench Hard | — | 56.1% |
| Epoch Capabilities Index | — | 111.97 |
| HellaSwag | — | 81% |
| PIQA | — | 83% |
| WinoGrande | — | 75.3% |
Math Not comparable
Codestral: —, Mistral 7B: 7.6 (#319)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 0.3% |
| LMArena Math | — | 1085 |
| MATH Level 5 | — | 3.7% |
| GSM8K | — | 54.4% |
Knowledge Not comparable
Codestral: —, Mistral 7B: 7.3 (#306)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| GPQA Diamond | — | 15.2% |
| LMArena Expert | — | 1036 |
| ARC (AI2) Challenge | — | 78.6% |
| BoolQ | — | 87.4% |
| MMLU | — | 62.5% |
| OpenBookQA | — | 79.8% |
| TriviaQA | — | 75.2% |
Multilingual Not comparable
Codestral: —, Mistral 7B: 26.0 (#278)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| LMArena Non-English | — | 1012 |
| LMArena Chinese | — | 1009 |
| LMArena French | — | 1036 |
| LMArena German | — | 987 |
| LMArena Japanese | — | 878 |
| LMArena Russian | — | 1018 |
| LMArena Spanish | — | 1026 |
Instruction Following Not comparable
Codestral: —, Mistral 7B: 54.5 (#275)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| LMArena Instruction Following | — | 1060 |
Long Context Not comparable
Codestral: —, Mistral 7B: 32.2 (#266)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| LMArena Longer Query | — | 1060 |
Writing & Preference Not comparable
Codestral: —, Mistral 7B: 30.9 (#281)
| Benchmark | Codestral | Mistral 7B |
|---|---|---|
| LMArena Text | — | 1090 |
| LMArena Creative Writing | — | 1068 |
| LMArena Multi-Turn | — | 1062 |
Frequently asked questions
Is Codestral better than Mistral 7B?
Codestral is the stronger model overall, scoring 30.1 to 22.9 on the Noometry Index. Mistral 7B costs 1.8× less per token, which makes it the better buy when Codestral's lead doesn't matter for your workload.
Which is cheaper, Codestral or Mistral 7B?
Mistral 7B is cheaper. It lists at $0.25 per million input tokens and $0.25 per million output tokens; Codestral lists at $0.30 and $0.90.
Is Codestral or Mistral 7B better for coding?
They score almost the same on coding (26.9 vs 25.9); test both on your own repository before choosing.
Which has the bigger context window?
Codestral does, with 256K tokens against 8K.
How many benchmarks do Codestral and Mistral 7B share?
4 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and Mistral 7B has 37.