Model comparison
Llama 3.1-8B vs Mixtral 8x22B
Mixtral 8x22B is the stronger model overall, scoring 27.1 to 23.0 on the Noometry Index. Llama 3.1-8B costs 52× less per token, which makes it the better buy when Mixtral 8x22B's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
Meta
Mistral AISummary
- They share 32 benchmarks with published results for both. Llama 3.1-8B scores higher in 3 categories and Mixtral 8x22B in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Mixtral 8x22B leads 22.9 to 10.2.
- The biggest single-benchmark swing is BigCodeBench Complete: 40.5% for Llama 3.1-8B and 50.2% for Mixtral 8x22B.
- Llama 3.1-8B is cheaper at $0.05 / $0.08 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- Llama 3.1-8B accepts more context: 128K tokens versus 64K.
Side by side
| Llama 3.1-8B | Mixtral 8x22B | |
|---|---|---|
| Provider | Meta | Mistral AI |
| Noometry Index | 23.0 | 27.1 |
| Released | 2024-07-23 | 2024-04-17 |
| Weights | Open | Open |
| Context window | 128K | 64K |
| Max output | 4K | 64K |
| Input $ / M tokens | $0.05 | $2 |
| Output $ / M tokens | $0.08 | $6 |
| Results tracked | 43 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Mixtral 8x22B leads
Llama 3.1-8B: 20.2 (#340), Mixtral 8x22B: 24.2 (#329)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| WeirdML | 1.7% | 3.2% |
| BigCodeBench Instruct | 32.8% | 40.6% |
| LMArena Coding | 1195 | 1166 |
| BigCodeBench Complete | 40.5% | 50.2% |
| HumanEval+ | 62.8% | 72% |
| MBPP+ | 55.6% | 64.3% |
| SciCode | 13.2% | — |
Agentic & Tool Use Too close to call
Llama 3.1-8B: 22.5 (#131), Mixtral 8x22B: 23.1 (#127)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 25.8% | — |
| Cybench | — | 7.5% |
| BALROG | 15.1% | — |
Reasoning Mixtral 8x22B leads
Llama 3.1-8B: 14.9 (#321), Mixtral 8x22B: 19.9 (#248)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| LMArena Hard Prompts | 1175 | 1150 |
| DTBench | 50.9% | 55.1% |
| Epoch Capabilities Index | 116.57 | 122.03 |
| CritPt | 0% | — |
| Chess Puzzles | 0% | — |
| LMCA | 5.4% | — |
| ForecastBench | — | 56.3 |
| PIQA | 81.2% | — |
Math Mixtral 8x22B leads
Llama 3.1-8B: 10.2 (#317), Mixtral 8x22B: 22.9 (#275)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| Omni-MATH | 13.7% | 16.3% |
| LMArena Math | 1179 | 1184 |
| MATH Level 5 | 22.9% | 24.2% |
| OTIS Mock AIME 2024-2025 | 1.7% | — |
| GSM8K | 82.4% | — |
Knowledge Mixtral 8x22B leads
Llama 3.1-8B: 8.0 (#307), Mixtral 8x22B: 15.1 (#293)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| GPQA Diamond | 27% | 34.1% |
| MMLU-Pro | 40.6% | 46% |
| GPQA (HELM) | 24.7% | 33.4% |
| LMArena Expert | 1144 | 1113 |
| MMLU | 56.1% | 77.8% |
| BoolQ | 82.8% | — |
Multilingual Llama 3.1-8B leads
Llama 3.1-8B: 34.0 (#249), Mixtral 8x22B: 32.8 (#255)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| LMArena Non-English | 1148 | 1128 |
| LMArena Chinese | 1151 | 1116 |
| LMArena French | 1177 | 1166 |
| LMArena German | 1144 | 1141 |
| LMArena Japanese | 1061 | 1037 |
| LMArena Korean | 1053 | 1057 |
| LMArena Russian | 1158 | 1158 |
| LMArena Spanish | 1169 | 1151 |
Instruction Following Llama 3.1-8B leads
Llama 3.1-8B: 58.9 (#258), Mixtral 8x22B: 57.7 (#266)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| IFEval | 74.3% | 72.4% |
| LMArena Instruction Following | 1159 | 1147 |
Long Context Llama 3.1-8B leads
Llama 3.1-8B: 35.8 (#238), Mixtral 8x22B: 34.7 (#247)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| LMArena Longer Query | 1182 | 1144 |
Writing & Preference Mixtral 8x22B leads
Llama 3.1-8B: 29.7 (#290), Mixtral 8x22B: 36.9 (#262)
| Benchmark | Llama 3.1-8B | Mixtral 8x22B |
|---|---|---|
| LMArena Text | 1187 | 1162 |
| LMArena Creative Writing | 1154 | 1141 |
| WildBench | 68.7% | 71.1% |
| LMArena Multi-Turn | 1172 | 1130 |
| EQ-Bench Creative Writing | 713 | — |
Frequently asked questions
Is Llama 3.1-8B better than Mixtral 8x22B?
Mixtral 8x22B is the stronger model overall, scoring 27.1 to 23.0 on the Noometry Index. Llama 3.1-8B costs 52× less per token, which makes it the better buy when Mixtral 8x22B's lead doesn't matter for your workload.
Which is cheaper, Llama 3.1-8B or Mixtral 8x22B?
Llama 3.1-8B is cheaper. It lists at $0.05 per million input tokens and $0.08 per million output tokens; Mixtral 8x22B lists at $2 and $6.
Is Llama 3.1-8B or Mixtral 8x22B better for coding?
Mixtral 8x22B scores higher on coding benchmarks: 24.2 versus 20.2 in the Noometry coding category.
Which has the bigger context window?
Llama 3.1-8B does, with 128K tokens against 64K.
How many benchmarks do Llama 3.1-8B and Mixtral 8x22B share?
32 benchmarks have published results for both models. Llama 3.1-8B has 43 scored results on Noometry and Mixtral 8x22B has 34.