Model comparison
Llama-3.3-70B-Instruct vs Mixtral 8x22B
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.6 to 27.1 on the Noometry Index.
Last verified . 26 shared benchmarks.
Meta
Mistral AISummary
- They share 26 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 6 categories and Mixtral 8x22B in 3 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Llama-3.3-70B-Instruct leads 30.6 to 15.1.
- The biggest single-benchmark swing is MATH Level 5: 41.6% for Llama-3.3-70B-Instruct and 24.2% for Mixtral 8x22B.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- Llama-3.3-70B-Instruct accepts more context: 128K tokens versus 64K.
Side by side
| Llama-3.3-70B-Instruct | Mixtral 8x22B | |
|---|---|---|
| Provider | Meta | Mistral AI |
| Noometry Index | 30.6 | 27.1 |
| Released | 2024-12-06 | 2024-04-17 |
| Weights | Open | Open |
| Context window | 128K | 64K |
| Max output | 4K | 64K |
| Input $ / M tokens | $0.10 | $2 |
| Output $ / M tokens | $0.32 | $6 |
| Results tracked | 43 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 31.0 (#290), Mixtral 8x22B: 24.2 (#329)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| WeirdML | 14.4% | 3.2% |
| BigCodeBench Instruct | 46.9% | 40.6% |
| LMArena Coding | 1268 | 1166 |
| BigCodeBench Complete | 57.5% | 50.2% |
| SciCode | 26% | — |
| LiveBench Coding | 36.6% | — |
| HumanEval+ | — | 72% |
| MBPP+ | — | 64.3% |
Agentic & Tool Use Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 25.8 (#105), Mixtral 8x22B: 23.1 (#127)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| Berkeley Function Calling Leaderboard | 31.9% | — |
| Cybench | — | 7.5% |
| BALROG | 23% | — |
Reasoning Mixtral 8x22B leads
Llama-3.3-70B-Instruct: 14.1 (#327), Mixtral 8x22B: 19.9 (#248)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Hard Prompts | 1257 | 1150 |
| DTBench | 59.5% | 55.1% |
| Epoch Capabilities Index | 127.33 | 122.03 |
| ForecastBench | 58.6 | 56.3 |
| SimpleBench | 19.9% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 50.8% | — |
| LiveBench Data Analysis | 49.5% | — |
| LMCA | 17.5% | — |
| LiveBench | 50.2% | — |
Math Mixtral 8x22B leads
Llama-3.3-70B-Instruct: 15.3 (#298), Mixtral 8x22B: 22.9 (#275)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Math | 1267 | 1184 |
| MATH Level 5 | 41.6% | 24.2% |
| OTIS Mock AIME 2024-2025 | 5.1% | — |
| Omni-MATH | — | 16.3% |
| LiveBench Math | 42.2% | — |
Knowledge Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 30.6 (#226), Mixtral 8x22B: 15.1 (#293)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| GPQA Diamond | 47.4% | 34.1% |
| LMArena Expert | 1225 | 1113 |
| MMLU | 86.3% | 77.8% |
| MMLU-Pro | — | 46% |
| Confabulations | 22.8% | — |
| Vectara Hallucination Rate | 4.1% | — |
| GPQA (HELM) | — | 33.4% |
Multilingual Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 39.9 (#220), Mixtral 8x22B: 32.8 (#255)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Non-English | 1236 | 1128 |
| LMArena Chinese | 1217 | 1116 |
| LMArena French | 1281 | 1166 |
| LMArena German | 1251 | 1141 |
| LMArena Japanese | 1150 | 1037 |
| LMArena Korean | 1143 | 1057 |
| LMArena Russian | 1252 | 1158 |
| LMArena Spanish | 1270 | 1151 |
Instruction Following Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 71.1 (#157), Mixtral 8x22B: 57.7 (#266)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Instruction Following | 1242 | 1147 |
| LiveBench Instruction Following | 82.7% | — |
| IFEval | — | 72.4% |
Long Context Mixtral 8x22B leads
Llama-3.3-70B-Instruct: 26.4 (#295), Mixtral 8x22B: 34.7 (#247)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Longer Query | 1256 | 1144 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference Llama-3.3-70B-Instruct leads
Llama-3.3-70B-Instruct: 47.6 (#207), Mixtral 8x22B: 36.9 (#262)
| Benchmark | Llama-3.3-70B-Instruct | Mixtral 8x22B |
|---|---|---|
| LMArena Text | 1274 | 1162 |
| LMArena Creative Writing | 1250 | 1141 |
| LMArena Multi-Turn | 1280 | 1130 |
| WildBench | — | 71.1% |
| LiveBench Language | 39.2% | — |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than Mixtral 8x22B?
Llama-3.3-70B-Instruct is the stronger model overall, scoring 30.6 to 27.1 on the Noometry Index.
Which is cheaper, Llama-3.3-70B-Instruct or Mixtral 8x22B?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; Mixtral 8x22B lists at $2 and $6.
Is Llama-3.3-70B-Instruct or Mixtral 8x22B better for coding?
Llama-3.3-70B-Instruct scores higher on coding benchmarks: 31.0 versus 24.2 in the Noometry coding category.
Which has the bigger context window?
Llama-3.3-70B-Instruct does, with 128K tokens against 64K.
How many benchmarks do Llama-3.3-70B-Instruct and Mixtral 8x22B share?
26 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and Mixtral 8x22B has 34.