Model comparison
Codestral vs GPT-4
Codestral is the stronger model overall, scoring 30.6 to 29.1 on the Noometry Index.
Last verified . 3 shared benchmarks.
Mistral AISummary
- They share 3 benchmarks with published results for both. Codestral scores higher in 1 category and GPT-4 in 1 category; 2 gaps are clear of the uncertainty.
- The widest gap is in coding, where GPT-4 leads 31.6 to 27.3.
- Codestral is cheaper at $0.30 / $0.90 per million input/output tokens, against $30 / $60 for GPT-4.
- Codestral accepts more context: 256K tokens versus 8K.
Side by side
| Codestral | GPT-4 | |
|---|---|---|
| Provider | Mistral AI | OpenAI |
| Noometry Index | 30.6 | 29.1 |
| Released | 2024-05-29 | 2023-03-14 |
| Weights | Proprietary | Proprietary |
| Context window | 256K | 8K |
| Max output | 8K | 8K |
| Input $ / M tokens | $0.30 | $30 |
| Output $ / M tokens | $0.90 | $60 |
| Results tracked | 7 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-4 leads
Codestral: 27.3 (#321), GPT-4: 31.6 (#283)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| BigCodeBench Instruct | 41.8% | 46% |
| BigCodeBench Complete | 52.5% | 57.2% |
| HumanEval+ | 73.8% | 79.3% |
| Aider Polyglot | 11.1% | — |
| WeirdML | — | 12.4% |
| LMArena Coding | — | 1254 |
| ALE-Bench | 137.78 | — |
| MBPP+ | 61.9% | — |
Agentic & Tool Use Not comparable
Codestral: —, GPT-4: —
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| METR Time Horizons | — | 36.1% |
Reasoning Codestral leads
Codestral: 19.8 (#251), GPT-4: 17.8 (#289)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| Kagi LLM Benchmark | 32.5% | — |
| Chess Puzzles | — | 4% |
| LMArena Hard Prompts | — | 1241 |
| Mystery Game Puzzles | — | 12% |
| DTBench | — | 62.7% |
| LMCA | — | 17.1% |
| BIG-Bench Hard | — | 75.1% |
| Epoch Capabilities Index | — | 125.89 |
| ForecastBench | — | 57.8 |
| HellaSwag | — | 95.3% |
| WinoGrande | — | 87.5% |
Math Not comparable
Codestral: —, GPT-4: 10.8 (#309)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | — | 1.1% |
| LMArena Math | — | 1269 |
| MATH Level 5 | — | 23% |
| GSM8K | — | 92% |
Knowledge Not comparable
Codestral: —, GPT-4: 18.4 (#282)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| GPQA Diamond | — | 35.7% |
| LMArena Expert | — | 1211 |
| MMLU | — | 86.4% |
| TriviaQA | — | 84.8% |
Multilingual Not comparable
Codestral: —, GPT-4: 40.6 (#215)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| LMArena Non-English | — | 1246 |
| LMArena Chinese | — | 1242 |
| LMArena French | — | 1283 |
| LMArena German | — | 1251 |
| LMArena Japanese | — | 1209 |
| LMArena Korean | — | 1184 |
| LMArena Russian | — | 1251 |
| LMArena Spanish | — | 1261 |
Instruction Following Not comparable
Codestral: —, GPT-4: 65.3 (#222)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| LMArena Instruction Following | — | 1241 |
Long Context Not comparable
Codestral: —, GPT-4: 37.7 (#212)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| LMArena Longer Query | — | 1244 |
Writing & Preference Not comparable
Codestral: —, GPT-4: 34.9 (#268)
| Benchmark | Codestral | GPT-4 |
|---|---|---|
| LMArena Text | — | 1263 |
| LMArena Creative Writing | — | 1244 |
| EQ-Bench Creative Writing | — | 752 |
| LMArena Multi-Turn | — | 1257 |
Frequently asked questions
Is Codestral better than GPT-4?
Codestral is the stronger model overall, scoring 30.6 to 29.1 on the Noometry Index.
Which is cheaper, Codestral or GPT-4?
Codestral is cheaper. It lists at $0.30 per million input tokens and $0.90 per million output tokens; GPT-4 lists at $30 and $60.
Is Codestral or GPT-4 better for coding?
GPT-4 scores higher on coding benchmarks: 31.6 versus 27.3 in the Noometry coding category.
Which has the bigger context window?
Codestral does, with 256K tokens against 8K.
How many benchmarks do Codestral and GPT-4 share?
3 benchmarks have published results for both models. Codestral has 7 scored results on Noometry and GPT-4 has 38.