Model comparison
Mistral 7B vs Mixtral 8x7B
Mixtral 8x7B is the stronger model overall, scoring 27.1 to 23.0 on the Noometry Index. Mistral 7B costs 2.8× less per token, which makes it the better buy when Mixtral 8x7B's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
Mistral AISummary
- They share 31 benchmarks with published results for both. Mistral 7B scores higher in 1 category and Mixtral 8x7B in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Mixtral 8x7B leads 18.8 to 8.1.
- The biggest single-benchmark swing is GPQA Diamond: 15.2% for Mistral 7B and 30.6% for Mixtral 8x7B.
- Mistral 7B is cheaper at $0.25 / $0.25 per million input/output tokens, against $0.70 / $0.70 for Mixtral 8x7B.
- Mixtral 8x7B accepts more context: 32K tokens versus 8K.
Side by side
| Mistral 7B | Mixtral 8x7B | |
|---|---|---|
| Provider | Mistral AI | Mistral AI |
| Noometry Index | 23.0 | 27.1 |
| Released | 2023-09-27 | 2023-12-11 |
| Weights | Open | Open |
| Context window | 8K | 32K |
| Max output | 8K | 32K |
| Input $ / M tokens | $0.25 | $0.70 |
| Output $ / M tokens | $0.25 | $0.70 |
| Results tracked | 37 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mixtral 8x7B leads
Mistral 7B: 26.4 (#326), Mixtral 8x7B: 32.8 (#269)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Coding | 1082 | 1126 |
| HumanEval+ | 36% | 39.6% |
| MBPP+ | 42.1% | 49.7% |
| BigCodeBench Instruct | 19.5% | — |
| BigCodeBench Complete | 27.3% | — |
Reasoning Mixtral 8x7B leads
Mistral 7B: 13.1 (#336), Mixtral 8x7B: 18.2 (#285)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Hard Prompts | 1067 | 1115 |
| DTBench | 42.5% | 49.6% |
| Adversarial NLI | 47.1% | 55.2% |
| Epoch Capabilities Index | 112.21 | 118.47 |
| HellaSwag | 81% | 86.7% |
| PIQA | 83% | 83.6% |
| WinoGrande | 75.3% | 77.2% |
| Chess Puzzles | 0% | — |
| BIG-Bench Hard | 56.1% | — |
| ForecastBench | — | 56.3 |
Math Mixtral 8x7B leads
Mistral 7B: 8.1 (#325), Mixtral 8x7B: 18.8 (#289)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Math | 1085 | 1147 |
| MATH Level 5 | 3.7% | 10% |
| GSM8K | 54.4% | 74.4% |
| OTIS Mock AIME 2024-2025 | 0.3% | — |
| Omni-MATH | — | 10.5% |
Knowledge Mixtral 8x7B leads
Mistral 7B: 7.4 (#311), Mixtral 8x7B: 11.0 (#301)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| GPQA Diamond | 15.2% | 30.6% |
| LMArena Expert | 1036 | 1088 |
| ARC (AI2) Challenge | 78.6% | 87.3% |
| MMLU | 62.5% | 70.6% |
| OpenBookQA | 79.8% | 85.8% |
| TriviaQA | 75.2% | 82.2% |
| MMLU-Pro | — | 33.5% |
| GPQA (HELM) | — | 29.6% |
| BoolQ | 87.4% | — |
Multilingual Mixtral 8x7B leads
Mistral 7B: 25.8 (#283), Mixtral 8x7B: 29.6 (#266)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Non-English | 1012 | 1077 |
| LMArena Chinese | 1009 | 1055 |
| LMArena French | 1037 | 1166 |
| LMArena German | 987 | 1114 |
| LMArena Japanese | 878 | 931 |
| LMArena Russian | 1018 | 1090 |
| LMArena Spanish | 1026 | 1111 |
| LMArena Korean | — | 968 |
Instruction Following Mistral 7B leads
Mistral 7B: 54.2 (#280), Mixtral 8x7B: 51.0 (#297)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Instruction Following | 1060 | 1109 |
| IFEval | — | 57.5% |
Long Context Mixtral 8x7B leads
Mistral 7B: 32.2 (#271), Mixtral 8x7B: 33.4 (#260)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Longer Query | 1060 | 1103 |
Writing & Preference Mixtral 8x7B leads
Mistral 7B: 30.7 (#286), Mixtral 8x7B: 34.2 (#270)
| Benchmark | Mistral 7B | Mixtral 8x7B |
|---|---|---|
| LMArena Text | 1090 | 1132 |
| LMArena Creative Writing | 1068 | 1109 |
| LMArena Multi-Turn | 1062 | 1115 |
| WildBench | — | 67.3% |
Frequently asked questions
Is Mistral 7B better than Mixtral 8x7B?
Mixtral 8x7B is the stronger model overall, scoring 27.1 to 23.0 on the Noometry Index. Mistral 7B costs 2.8× less per token, which makes it the better buy when Mixtral 8x7B's lead doesn't matter for your workload.
Which is cheaper, Mistral 7B or Mixtral 8x7B?
Mistral 7B is cheaper. It lists at $0.25 per million input tokens and $0.25 per million output tokens; Mixtral 8x7B lists at $0.70 and $0.70.
Is Mistral 7B or Mixtral 8x7B better for coding?
Mixtral 8x7B scores higher on coding benchmarks: 32.8 versus 26.4 in the Noometry coding category.
Which has the bigger context window?
Mixtral 8x7B does, with 32K tokens against 8K.
How many benchmarks do Mistral 7B and Mixtral 8x7B share?
31 benchmarks have published results for both models. Mistral 7B has 37 scored results on Noometry and Mixtral 8x7B has 38.