Model comparison
Codestral vs Mistral Large
Mistral Large is the stronger model overall, scoring 31.6 to 30.1 on the Noometry Index. Codestral costs 6.7× less per token, which makes it the better buy when Mistral Large's lead doesn't matter for your workload.
Last verified . 5 shared benchmarks.
Mistral AISummary
- They share 5 benchmarks with published results for both. Codestral scores higher in 1 category and Mistral Large in 1 category; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where Mistral Large leads 33.8 to 26.9.
- The biggest single-benchmark swing is BigCodeBench Complete: 52.5% for Codestral and 38.3% for Mistral Large.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $2 / $6 for Mistral Large.
- Codestral accepts more context: 256K tokens versus 131K.
Side by side
| Codestral | Mistral Large | |
|---|---|---|
| Provider | Mistral AI | Mistral AI |
| Noometry Index | 30.1 | 31.6 |
| Released | 2024-05-29 | 2024-02-26 |
| Weights | Open | Open |
| Context window | 256K | 131K |
| Max output | 4K | 16K |
| Input $ / M tokens | $0.30 | $2 |
| Output $ / M tokens | $0.90 | $6 |
| Results tracked | 7 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Large leads
Codestral: 26.9 (#316), Mistral Large: 33.8 (#240)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| BigCodeBench Instruct | 41.8% | 30% |
| BigCodeBench Complete | 52.5% | 38.3% |
| ALE-Bench | 137.78 | 264.7 |
| HumanEval+ | 73.8% | 62.2% |
| MBPP+ | 61.9% | 59.5% |
| Aider Polyglot | 11.1% | — |
| SciCode | — | 36.2% |
| LiveBench Coding | — | 47.1% |
| LMArena Coding | — | 1277 |
Agentic & Tool Use Not comparable
Codestral: —, Mistral Large: 28.4 (#83)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 38.4% |
Reasoning Codestral leads
Codestral: 19.1 (#244), Mistral Large: 15.4 (#305)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| SimpleBench | — | 22.5% |
| Kagi LLM Benchmark | 32.5% | — |
| CritPt | — | 0% |
| LiveBench Reasoning | — | 43.5% |
| LMArena Hard Prompts | — | 1257 |
| DTBench | — | 65.1% |
| LiveBench Data Analysis | — | 50.1% |
| LMCA | — | 16.7% |
| Epoch Capabilities Index | — | 128.52 |
| ForecastBench | — | 57.1 |
| LiveBench | — | 48.4% |
Math Not comparable
Codestral: —, Mistral Large: 17.1 (#285)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 8.5% |
| Omni-MATH | — | 28.1% |
| LiveBench Math | — | 42.5% |
| LMArena Math | — | 1262 |
| MATH Level 5 | — | 50.3% |
| FrontierMath (Feb 2025 set) | — | 0.3% |
Knowledge Not comparable
Codestral: —, Mistral Large: 30.0 (#224)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| GPQA Diamond | — | 51.3% |
| MMLU-Pro | — | 59.9% |
| Confabulations | — | 21.4% |
| Vectara Hallucination Rate | — | 4.5% |
| GPQA (HELM) | — | 43.5% |
| LMArena Expert | — | 1232 |
| MMLU | — | 80% |
Multilingual Not comparable
Codestral: —, Mistral Large: 40.1 (#214)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| LMArena Non-English | — | 1237 |
| LMArena Chinese | — | 1240 |
| LMArena French | — | 1324 |
| LMArena German | — | 1254 |
| LMArena Japanese | — | 1187 |
| LMArena Korean | — | 1202 |
| LMArena Russian | — | 1257 |
| LMArena Spanish | — | 1268 |
Instruction Following Not comparable
Codestral: —, Mistral Large: 68.1 (#186)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| LiveBench Instruction Following | — | 67.9% |
| IFEval | — | 87.7% |
| LMArena Instruction Following | — | 1249 |
Long Context Not comparable
Codestral: —, Mistral Large: 38.3 (#194)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| LMArena Longer Query | — | 1262 |
Writing & Preference Not comparable
Codestral: —, Mistral Large: 40.9 (#237)
| Benchmark | Codestral | Mistral Large |
|---|---|---|
| LMArena Text | — | 1266 |
| LMArena Creative Writing | — | 1243 |
| Short-Story Creative Writing | — | 69% |
| EQ-Bench Creative Writing | — | 985 |
| WildBench | — | 80.1% |
| LMArena Multi-Turn | — | 1260 |
| LiveBench Language | — | 39.4% |
Frequently asked questions
Is Codestral better than Mistral Large?
Mistral Large is the stronger model overall, scoring 31.6 to 30.1 on the Noometry Index. Codestral costs 6.7× less per token, which makes it the better buy when Mistral Large's lead doesn't matter for your workload.
Which is cheaper, Codestral or Mistral Large?
Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; Mistral Large lists at $2 and $6.
Is Codestral or Mistral Large better for coding?
Mistral Large scores higher on coding benchmarks: 33.8 versus 26.9 in the Noometry coding category.
Which has the bigger context window?
Codestral does, with 256K tokens against 131K.
How many benchmarks do Codestral and Mistral Large share?
5 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and Mistral Large has 51.