Model comparison
GPT-4 vs Mistral Small 3.1
Mistral Small 3.1 is the stronger model overall, scoring 31.7 to 29.1 on the Noometry Index.
Last verified . 22 shared benchmarks.
Mistral AISummary
- They share 22 benchmarks with published results for both. GPT-4 scores higher in 1 category and Mistral Small 3.1 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in coding, where Mistral Small 3.1 leads 38.3 to 31.6.
- The biggest single-benchmark swing is GPQA Diamond: 35.7% for GPT-4 and 41.9% for Mistral Small 3.1.
- Mistral Small 3.1 is cheaper at $0.35 / $0.56 per million input/output tokens, against $30 / $60 for GPT-4.
- Mistral Small 3.1 accepts more context: 128K tokens versus 8K.
- Mistral Small 3.1 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Mistral Small 3.1 | |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Noometry Index | 29.1 | 31.7 |
| Released | 2023-03-14 | 2025-03-17 |
| Weights | Proprietary | Open |
| Context window | 8K | 128K |
| Max output | 8K | 102K |
| Input $ / M tokens | $30 | $0.35 |
| Output $ / M tokens | $60 | $0.56 |
| Results tracked | 38 | 28 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mistral Small 3.1 leads
GPT-4: 31.6 (#283), Mistral Small 3.1: 38.3 (#179)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Coding | 1254 | 1309 |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Mistral Small 3.1: —
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Mistral Small 3.1 leads
GPT-4: 17.8 (#289), Mistral Small 3.1: 19.7 (#254)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| Chess Puzzles | 4% | 1% |
| LMArena Hard Prompts | 1241 | 1278 |
| Epoch Capabilities Index | 125.89 | 127.48 |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Mistral Small 3.1 leads
GPT-4: 10.8 (#309), Mistral Small 3.1: 14.7 (#301)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 3.9% |
| LMArena Math | 1269 | 1262 |
| Omni-MATH | — | 24.8% |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Mistral Small 3.1 leads
GPT-4: 18.4 (#282), Mistral Small 3.1: 22.6 (#271)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| GPQA Diamond | 35.7% | 41.9% |
| LMArena Expert | 1211 | 1257 |
| MMLU-Pro | — | 61% |
| GPQA (HELM) | — | 39.2% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, Mistral Small 3.1: 33.2 (#99)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Vision | — | 1136 |
Multilingual Too close to call
GPT-4: 40.6 (#215), Mistral Small 3.1: 41.2 (#209)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Non-English | 1246 | 1255 |
| LMArena Chinese | 1242 | 1253 |
| LMArena French | 1283 | 1273 |
| LMArena German | 1251 | 1266 |
| LMArena Japanese | 1209 | 1208 |
| LMArena Korean | 1184 | 1206 |
| LMArena Russian | 1251 | 1263 |
| LMArena Spanish | 1261 | 1283 |
Instruction Following GPT-4 leads
GPT-4: 65.3 (#222), Mistral Small 3.1: 63.6 (#230)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1264 |
| IFEval | — | 75% |
Long Context Mistral Small 3.1 leads
GPT-4: 37.7 (#212), Mistral Small 3.1: 39.5 (#178)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Longer Query | 1244 | 1299 |
Writing & Preference Mistral Small 3.1 leads
GPT-4: 34.9 (#268), Mistral Small 3.1: 37.0 (#259)
| Benchmark | GPT-4 | Mistral Small 3.1 |
|---|---|---|
| LMArena Text | 1263 | 1277 |
| LMArena Creative Writing | 1244 | 1253 |
| EQ-Bench Creative Writing | 752 | 761 |
| LMArena Multi-Turn | 1257 | 1270 |
| WildBench | — | 78.8% |
Frequently asked questions
Is GPT-4 better than Mistral Small 3.1?
Mistral Small 3.1 is the stronger model overall, scoring 31.7 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 or Mistral Small 3.1?
Mistral Small 3.1 is cheaper. It lists at $0.35 per million input tokens and $0.56 per million output tokens; GPT-4 lists at $30 and $60.
Is GPT-4 or Mistral Small 3.1 better for coding?
Mistral Small 3.1 scores higher on coding benchmarks: 38.3 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
Mistral Small 3.1 does, with 128K tokens against 8K.
How many benchmarks do GPT-4 and Mistral Small 3.1 share?
22 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Mistral Small 3.1 has 28.