Model comparison
GPT-4 vs Ministral 3B
GPT-4 is the stronger model overall, scoring 29.1 to 26.2 on the Noometry Index. Ministral 3B costs 375× less per token, which makes it the better buy when GPT-4's lead doesn't matter for your workload.
Last verified . 5 shared benchmarks.
Mistral AISummary
- They share 5 benchmarks with published results for both. GPT-4 scores higher in 1 category and Ministral 3B in 2 categories; 2 gaps are clear of the uncertainty.
- The widest gap is in math, where Ministral 3B leads 26.6 to 10.8.
- The biggest single-benchmark swing is LMCA: 17.1% for GPT-4 and 5.5% for Ministral 3B.
- Ministral 3B is cheaper at $0.10 / $0.10 per million input/output tokens, against $30 / $60 for GPT-4.
- Ministral 3B accepts more context: 131K tokens versus 8K.
- Ministral 3B has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Ministral 3B | |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Noometry Index | 29.1 | 26.2 |
| Released | 2023-03-14 | 2024-10-01 |
| Weights | Proprietary | Open |
| Context window | 8K | 131K |
| Max output | 8K | 262K |
| Input $ / M tokens | $30 | $0.10 |
| Output $ / M tokens | $60 | $0.10 |
| Results tracked | 38 | 6 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
GPT-4: 31.6 (#283), Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| LMArena Coding | 1254 | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Too close to call
GPT-4: 17.8 (#289), Ministral 3B: 18.4 (#282)
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| DTBench | 62.7% | 51.7% |
| LMCA | 17.1% | 5.5% |
| Epoch Capabilities Index | 125.89 | 118.1 |
| Chess Puzzles | 4% | — |
| LMArena Hard Prompts | 1241 | — |
| Mystery Game Puzzles | 12% | — |
| BIG-Bench Hard | 75.1% | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Ministral 3B leads
GPT-4: 10.8 (#309), Ministral 3B: 26.6 (#258)
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| MATH Level 5 | 23% | 14.4% |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| LMArena Math | 1269 | — |
| GSM8K | 92% | — |
Knowledge GPT-4 leads
GPT-4: 18.4 (#282), Ministral 3B: 10.4 (#302)
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| GPQA Diamond | 35.7% | 25.3% |
| Vectara Hallucination Rate | — | 7.3% |
| LMArena Expert | 1211 | — |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multilingual Not comparable
GPT-4: 40.6 (#215), Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| 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
GPT-4: 65.3 (#222), Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| LMArena Instruction Following | 1241 | — |
Long Context Not comparable
GPT-4: 37.7 (#212), Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| LMArena Longer Query | 1244 | — |
Writing & Preference Not comparable
GPT-4: 34.9 (#268), Ministral 3B: —
| Benchmark | GPT-4 | Ministral 3B |
|---|---|---|
| LMArena Text | 1263 | — |
| LMArena Creative Writing | 1244 | — |
| EQ-Bench Creative Writing | 752 | — |
| LMArena Multi-Turn | 1257 | — |
Frequently asked questions
Is GPT-4 better than Ministral 3B?
GPT-4 is the stronger model overall, scoring 29.1 to 26.2 on the Noometry Index. Ministral 3B costs 375× less per token, which makes it the better buy when GPT-4's lead doesn't matter for your workload.
Which is cheaper, GPT-4 or Ministral 3B?
Ministral 3B is cheaper. It lists at $0.10 per million input tokens and $0.10 per million output tokens; GPT-4 lists at $30 and $60.
Which has the bigger context window?
Ministral 3B does, with 131K tokens against 8K.
How many benchmarks do GPT-4 and Ministral 3B share?
5 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Ministral 3B has 6.