Model comparison
Llama 3.1-8B vs Mixtral 8x7B
Mixtral 8x7B is the stronger model overall, scoring 26.8 to 22.8 on the Noometry Index. Llama 3.1-8B costs 12× 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.
Meta
Mistral AISummary
- They share 31 benchmarks with published results for both. Llama 3.1-8B scores higher in 3 categories and Mixtral 8x7B in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in coding, where Mixtral 8x7B leads 32.2 to 19.9.
- The biggest single-benchmark swing is IFEval: 74.3% for Llama 3.1-8B and 57.5% for Mixtral 8x7B.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $0.70 / $0.70 for Mixtral 8x7B.
- Llama 3.1-8B accepts more context: 128K tokens versus 32K.
Side by side
| Llama 3.1-8B | Mixtral 8x7B | |
|---|---|---|
| Provider | Meta | Mistral AI |
| Noometry Index | 22.8 | 26.8 |
| Released | 2024-07-23 | 2023-12-11 |
| Weights | Open | Open |
| Context window | 128K | 32K |
| Max output | 4K | 32K |
| Input $ / M tokens | $0.05 | $0.70 |
| Output $ / M tokens | $0.08 | $0.70 |
| Results tracked | 43 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mixtral 8x7B leads
Llama 3.1-8B: 19.9 (#335), Mixtral 8x7B: 32.2 (#265)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| LMArena Coding | 1195 | 1127 |
| HumanEval+ | 62.8% | 39.6% |
| MBPP+ | 55.6% | 49.7% |
| SciCode | 13.2% | — |
| WeirdML | 1.7% | — |
| BigCodeBench Instruct | 32.8% | — |
| BigCodeBench Complete | 40.5% | — |
Agentic & Tool Use Not comparable
Llama 3.1-8B: 22.3 (#123), Mixtral 8x7B: —
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25.8% | — |
| BALROG | 15.1% | — |
Reasoning Mixtral 8x7B leads
Llama 3.1-8B: 14.5 (#316), Mixtral 8x7B: 17.4 (#283)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| LMArena Hard Prompts | 1175 | 1115 |
| DTBench | 50.9% | 49.6% |
| Epoch Capabilities Index | 116.48 | 118.38 |
| PIQA | 81.2% | 83.6% |
| CritPt | 0% | — |
| Chess Puzzles | 0% | — |
| LMCA | 5.4% | — |
| Adversarial NLI | — | 55.2% |
| ForecastBench | — | 56.3 |
| HellaSwag | — | 86.7% |
| WinoGrande | — | 77.2% |
Math Mixtral 8x7B leads
Llama 3.1-8B: 9.5 (#311), Mixtral 8x7B: 17.2 (#284)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| Omni-MATH | 13.7% | 10.5% |
| LMArena Math | 1179 | 1147 |
| MATH Level 5 | 22.9% | 10% |
| GSM8K | 82.4% | 74.4% |
| OTIS Mock AIME 2024-2025 | 1.7% | — |
Knowledge Mixtral 8x7B leads
Llama 3.1-8B: 8.0 (#302), Mixtral 8x7B: 11.0 (#296)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| GPQA Diamond | 27% | 30.6% |
| MMLU-Pro | 40.6% | 33.5% |
| GPQA (HELM) | 24.7% | 29.6% |
| LMArena Expert | 1144 | 1088 |
| MMLU | 56.1% | 70.6% |
| ARC (AI2) Challenge | — | 87.3% |
| BoolQ | 82.8% | — |
| OpenBookQA | — | 85.8% |
| TriviaQA | — | 82.2% |
Multilingual Llama 3.1-8B leads
Llama 3.1-8B: 34.2 (#244), Mixtral 8x7B: 29.8 (#261)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| LMArena Non-English | 1148 | 1077 |
| LMArena Chinese | 1151 | 1055 |
| LMArena French | 1176 | 1165 |
| LMArena German | 1144 | 1114 |
| LMArena Japanese | 1061 | 931 |
| LMArena Korean | 1053 | 968 |
| LMArena Russian | 1158 | 1090 |
| LMArena Spanish | 1169 | 1110 |
Instruction Following Llama 3.1-8B leads
Llama 3.1-8B: 59.1 (#253), Mixtral 8x7B: 51.2 (#292)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| IFEval | 74.3% | 57.5% |
| LMArena Instruction Following | 1159 | 1110 |
Long Context Llama 3.1-8B leads
Llama 3.1-8B: 35.8 (#233), Mixtral 8x7B: 33.4 (#255)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| LMArena Longer Query | 1182 | 1103 |
Writing & Preference Mixtral 8x7B leads
Llama 3.1-8B: 29.9 (#285), Mixtral 8x7B: 34.4 (#265)
| Benchmark | Llama 3.1-8B | Mixtral 8x7B |
|---|---|---|
| LMArena Text | 1187 | 1132 |
| LMArena Creative Writing | 1154 | 1109 |
| WildBench | 68.7% | 67.3% |
| LMArena Multi-Turn | 1172 | 1115 |
| EQ-Bench Creative Writing | 713 | — |
Frequently asked questions
Is Llama 3.1-8B better than Mixtral 8x7B?
Mixtral 8x7B is the stronger model overall, scoring 26.8 to 22.8 on the Noometry Index. Llama 3.1-8B costs 12× less per token, which makes it the better buy when Mixtral 8x7B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-8B or Mixtral 8x7B?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; Mixtral 8x7B lists at $0.70 and $0.70.
Is Llama 3.1-8B or Mixtral 8x7B better for coding?
Mixtral 8x7B scores higher on coding benchmarks: 32.2 versus 19.9 in the Noometry coding category.
Which has the bigger context window?
Llama 3.1-8B does, with 128K tokens against 32K.
How many benchmarks do Llama 3.1-8B and Mixtral 8x7B share?
31 benchmarks have published results for both models. Llama 3.1-8B has 43 scored results on Noometry and Mixtral 8x7B has 38.