Model comparison
DeepSeek-V3 vs Mixtral 8x22B
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 27.1 on the Noometry Index.
Last verified . 33 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 33 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Mixtral 8x22B in 1 category; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3 leads 37.5 to 15.1.
- The biggest single-benchmark swing is MATH Level 5: 75.5% for DeepSeek-V3 and 24.2% for Mixtral 8x22B.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- DeepSeek-V3 accepts more context: 164K tokens versus 64K.
Side by side
| DeepSeek-V3 | Mixtral 8x22B | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 39.5 | 27.1 |
| Released | 2024-12-26 | 2024-04-17 |
| Weights | Open | Open |
| Context window | 164K | 64K |
| Max output | 164K | 64K |
| Input $ / M tokens | $0.24 | $2 |
| Output $ / M tokens | $0.90 | $6 |
| Results tracked | 60 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Mixtral 8x22B: 24.2 (#329)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| WeirdML | 36.1% | 3.2% |
| BigCodeBench Instruct | 50% | 40.6% |
| LMArena Coding | 1368 | 1166 |
| BigCodeBench Complete | 62.2% | 50.2% |
| HumanEval+ | 86.6% | 72% |
| MBPP+ | 73% | 64.3% |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| LiveBench Coding | 70.9% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Mixtral 8x22B: 23.1 (#127)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| Cybench | — | 7.5% |
| METR Time Horizons | 49.6% | — |
Reasoning Too close to call
DeepSeek-V3: 20.5 (#236), Mixtral 8x22B: 19.9 (#248)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| LMArena Hard Prompts | 1365 | 1150 |
| DTBench | 64.8% | 55.1% |
| Epoch Capabilities Index | 135.94 | 122.03 |
| ForecastBench | 59.1 | 56.3 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LiveBench Reasoning | 65.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Mixtral 8x22B: 22.9 (#275)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| Omni-MATH | 40.3% | 16.3% |
| LMArena Math | 1373 | 1184 |
| MATH Level 5 | 75.5% | 24.2% |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| LiveBench Math | 73.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Mixtral 8x22B: 15.1 (#293)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| GPQA Diamond | 67.6% | 34.1% |
| MMLU-Pro | 72.3% | 46% |
| GPQA (HELM) | 53.8% | 33.4% |
| LMArena Expert | 1351 | 1113 |
| MMLU | 87.2% | 77.8% |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| ARC (AI2) Challenge | 95.3% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Mixtral 8x22B: 32.8 (#255)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| LMArena Non-English | 1358 | 1128 |
| LMArena Chinese | 1391 | 1116 |
| LMArena French | 1385 | 1166 |
| LMArena German | 1374 | 1141 |
| LMArena Japanese | 1333 | 1037 |
| LMArena Korean | 1319 | 1057 |
| LMArena Russian | 1373 | 1158 |
| LMArena Spanish | 1358 | 1151 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Mixtral 8x22B: 57.7 (#266)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| IFEval | 83.2% | 72.4% |
| LMArena Instruction Following | 1345 | 1147 |
| LiveBench Instruction Following | 81.5% | — |
Long Context Too close to call
DeepSeek-V3: 34.0 (#253), Mixtral 8x22B: 34.7 (#247)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| LMArena Longer Query | 1352 | 1144 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Mixtral 8x22B: 36.9 (#262)
| Benchmark | DeepSeek-V3 | Mixtral 8x22B |
|---|---|---|
| LMArena Text | 1375 | 1162 |
| LMArena Creative Writing | 1364 | 1141 |
| WildBench | 83% | 71.1% |
| LMArena Multi-Turn | 1389 | 1130 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Mixtral 8x22B?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 27.1 on the Noometry Index.
Which is cheaper, DeepSeek-V3 or Mixtral 8x22B?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Mixtral 8x22B lists at $2 and $6.
Is DeepSeek-V3 or Mixtral 8x22B better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 24.2 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 64K.
How many benchmarks do DeepSeek-V3 and Mixtral 8x22B share?
33 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Mixtral 8x22B has 34.