Model comparison
GPT-4 Turbo vs Mistral Small
Mistral Small is the stronger model overall, scoring 33.4 to 30.5 on the Noometry Index.
Last verified . 26 shared benchmarks.
Mistral AISummary
- They share 26 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and Mistral Small in 9 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Mistral Small leads 16.4 to 9.0.
- The biggest single-benchmark swing is BigCodeBench Instruct: 48.2% for GPT-4 Turbo and 36.1% for Mistral Small.
- Mistral Small is cheaper at $0.15 / $0.60 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- Mistral Small accepts more context: 262K tokens versus 128K.
- Mistral Small has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | Mistral Small | |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Noometry Index | 30.5 | 33.4 |
| Released | 2023-11-06 | 2024-02-26 |
| Weights | Proprietary | Open |
| Context window | 128K | 262K |
| Max output | 4K | 256K |
| Input $ / M tokens | $10 | $0.15 |
| Output $ / M tokens | $30 | $0.60 |
| Results tracked | 36 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
GPT-4 Turbo: 33.8 (#249), Mistral Small: 34.0 (#247)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| BigCodeBench Instruct | 48.2% | 36.1% |
| LMArena Coding | 1268 | 1362 |
| BigCodeBench Complete | 58.2% | 46.6% |
| SciCode | — | 26.5% |
| WeirdML | 18% | — |
| LiveBench Coding | — | 36.2% |
| ALE-Bench | — | 497.62 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, Mistral Small: 28.1 (#93)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.1% |
| METR Time Horizons | 36.7% | — |
Reasoning Mistral Small leads
GPT-4 Turbo: 15.3 (#317), Mistral Small: 19.8 (#250)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Hard Prompts | 1251 | 1335 |
| DTBench | 61.6% | 70.9% |
| LMCA | 9.8% | 20.6% |
| SimpleBench | 25.1% | — |
| Kagi LLM Benchmark | — | 37.8% |
| CritPt | — | 0% |
| Chess Puzzles | 6% | — |
| LiveBench Reasoning | — | 44.8% |
| LiveBench Data Analysis | — | 53.7% |
| Epoch Capabilities Index | 127.25 | — |
| ForecastBench | 59.4 | — |
| LiveBench | — | 44% |
Math Mistral Small leads
GPT-4 Turbo: 9.0 (#322), Mistral Small: 16.4 (#293)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 6.7% | 5.8% |
| LMArena Math | 1272 | 1341 |
| MATH Level 5 | 46.7% | 46.8% |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| LiveBench Math | — | 39.9% |
Knowledge Mistral Small leads
GPT-4 Turbo: 24.3 (#268), Mistral Small: 31.0 (#222)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| GPQA Diamond | 46.6% | 47.5% |
| LMArena Expert | 1223 | 1291 |
| MMLU | 81.3% | 68.7% |
| Confabulations | 28.4% | — |
| Vectara Hallucination Rate | — | 5.1% |
Multimodal Mistral Small leads
GPT-4 Turbo: 30.6 (#110), Mistral Small: 33.5 (#96)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Vision | 1090 | 1142 |
Multilingual Mistral Small leads
GPT-4 Turbo: 40.5 (#216), Mistral Small: 45.5 (#169)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Non-English | 1245 | 1315 |
| LMArena Chinese | 1242 | 1340 |
| LMArena French | 1276 | 1337 |
| LMArena German | 1259 | 1340 |
| LMArena Japanese | 1194 | 1275 |
| LMArena Korean | 1187 | 1259 |
| LMArena Russian | 1259 | 1324 |
| LMArena Spanish | 1260 | 1346 |
Instruction Following Too close to call
GPT-4 Turbo: 65.8 (#216), Mistral Small: 66.4 (#209)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Instruction Following | 1249 | 1310 |
| LiveBench Instruction Following | — | 63.7% |
Long Context Mistral Small leads
GPT-4 Turbo: 38.0 (#206), Mistral Small: 40.4 (#156)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Longer Query | 1254 | 1327 |
Writing & Preference Mistral Small leads
GPT-4 Turbo: 47.7 (#206), Mistral Small: 52.5 (#171)
| Benchmark | GPT-4 Turbo | Mistral Small |
|---|---|---|
| LMArena Text | 1272 | 1338 |
| LMArena Creative Writing | 1269 | 1305 |
| LMArena Multi-Turn | 1267 | 1344 |
| LiveBench Language | — | 30.5% |
Frequently asked questions
Is GPT-4 Turbo better than Mistral Small?
Mistral Small is the stronger model overall, scoring 33.4 to 30.5 on the Noometry Index.
Which is cheaper, GPT-4 Turbo or Mistral Small?
Mistral Small is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is GPT-4 Turbo or Mistral Small better for coding?
They score almost the same on coding (33.8 vs 34.0); test both on your own repository before choosing.
Which has the bigger context window?
Mistral Small does, with 262K tokens against 128K.
How many benchmarks do GPT-4 Turbo and Mistral Small share?
26 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Mistral Small has 39.