Model comparison
Codestral vs GPT-5 Mini
GPT-5 Mini is the stronger model overall, scoring 41.7 to 30.1 on the Noometry Index. Codestral costs 1.5× less per token, which makes it the better buy when GPT-5 Mini's lead doesn't matter for your workload.
Last verified . 2 shared benchmarks.
Mistral AISummary
- They share 2 benchmarks with published results for both. Codestral scores higher in 0 categories and GPT-5 Mini in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where GPT-5 Mini leads 40.0 to 26.9.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 32.5% for Codestral and 70.3% for GPT-5 Mini.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $0.25 / $2 for GPT-5 Mini.
- GPT-5 Mini accepts more context: 400K tokens versus 256K.
- Codestral has downloadable open weights; the other is API-only.
Side by side
| Codestral | GPT-5 Mini | |
|---|---|---|
| Provider | Mistral AI | OpenAI |
| Noometry Index | 30.1 | 41.7 |
| Released | 2024-05-29 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 256K | 400K |
| Max output | 4K | 128K |
| Input $ / M tokens | $0.30 | $0.25 |
| Output $ / M tokens | $0.90 | $2 |
| Results tracked | 7 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 Mini leads
Codestral: 26.9 (#316), GPT-5 Mini: 40.0 (#134)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| ALE-Bench | 137.78 | 799.77 |
| SWE-bench Verified | — | 64.7% |
| SWE-bench Verified (bash only) | — | 59.8% |
| Aider Polyglot | 11.1% | — |
| SWE-bench Multilingual | — | 39.7% |
| SciCode | — | 39.2% |
| WeirdML | — | 52.7% |
| BigCodeBench Instruct | 41.8% | — |
| LMArena Coding | — | 1406 |
| BigCodeBench Complete | 52.5% | — |
| AlgoTune | — | 1.38 |
| HumanEval+ | 73.8% | — |
| MBPP+ | 61.9% | — |
Agentic & Tool Use Not comparable
Codestral: —, GPT-5 Mini: 31.0 (#64)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| Terminal-Bench | — | 34.8% |
| Berkeley Function Calling Leaderboard | — | 55.5% |
| Vending-Bench 2 | — | -31.18 |
Reasoning GPT-5 Mini leads
Codestral: 19.1 (#244), GPT-5 Mini: 24.1 (#153)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | 70.3% |
| ARC-AGI-2 | — | 4.4% |
| ARC-AGI-1 | — | 54.3% |
| CritPt | — | 0% |
| Chess Puzzles | — | 30% |
| EnigmaEval | — | 8.2% |
| LMArena Hard Prompts | — | 1380 |
| Mystery Game Puzzles | — | 10% |
| DTBench | — | 80.5% |
| LMCA | — | 34.2% |
| Epoch Capabilities Index | — | 145.51 |
| ForecastBench | — | 61 |
Math Not comparable
Codestral: —, GPT-5 Mini: 46.3 (#59)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | — | 46.7% |
| FrontierMath Tier 4 | — | 12.2% |
| OTIS Mock AIME 2024-2025 | — | 86.7% |
| ProofBench | — | 9% |
| Omni-MATH | — | 72.2% |
| LMArena Math | — | 1377 |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 27.2% |
| FrontierMath Tier 4 (v1) | — | 6.3% |
Knowledge Not comparable
Codestral: —, GPT-5 Mini: 45.5 (#80)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| GPQA Diamond | — | 75% |
| Humanity's Last Exam | — | 19.4% |
| SimpleQA Verified | — | 21.6% |
| MMLU-Pro | — | 83.5% |
| Confabulations | — | 13.3% |
| Vectara Hallucination Rate | — | 12.9% |
| GPQA (HELM) | — | 75.6% |
| LMArena Expert | — | 1377 |
Multimodal Not comparable
Codestral: —, GPT-5 Mini: 35.0 (#80)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| LMArena Vision | — | 1202 |
| VPCT | — | 40.2% |
Multilingual Not comparable
Codestral: —, GPT-5 Mini: 49.1 (#131)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| LMArena Non-English | — | 1363 |
| LMArena Chinese | — | 1387 |
| LMArena French | — | 1385 |
| LMArena German | — | 1366 |
| LMArena Japanese | — | 1342 |
| LMArena Korean | — | 1306 |
| LMArena Russian | — | 1362 |
| LMArena Spanish | — | 1355 |
Instruction Following Not comparable
Codestral: —, GPT-5 Mini: 76.4 (#43)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| IFEval | — | 92.7% |
| LMArena Instruction Following | — | 1357 |
Long Context Not comparable
Codestral: —, GPT-5 Mini: 41.9 (#127)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| Fiction.LiveBench | — | 69.4% |
| LMArena Longer Query | — | 1355 |
Writing & Preference Not comparable
Codestral: —, GPT-5 Mini: 55.4 (#143)
| Benchmark | Codestral | GPT-5 Mini |
|---|---|---|
| LMArena Text | — | 1373 |
| LMArena Creative Writing | — | 1325 |
| Short-Story Creative Writing | — | 83.1% |
| EQ-Bench Creative Writing | — | 1313 |
| WildBench | — | 85.5% |
| LMArena Multi-Turn | — | 1363 |
Frequently asked questions
Is Codestral better than GPT-5 Mini?
GPT-5 Mini is the stronger model overall, scoring 41.7 to 30.1 on the Noometry Index. Codestral costs 1.5× less per token, which makes it the better buy when GPT-5 Mini's lead doesn't matter for your workload.
Which is cheaper, Codestral or GPT-5 Mini?
Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; GPT-5 Mini lists at $0.25 and $2.
Is Codestral or GPT-5 Mini better for coding?
GPT-5 Mini scores higher on coding benchmarks: 40.0 versus 26.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Mini does, with 400K tokens against 256K.
How many benchmarks do Codestral and GPT-5 Mini share?
2 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and GPT-5 Mini has 60.