Model comparison
DeepSeek-R1 vs Mixtral 8x22B
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 27.1 on the Noometry Index.
Last verified . 27 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 27 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and Mixtral 8x22B in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-R1 leads 44.5 to 15.1.
- The biggest single-benchmark swing is MATH Level 5: 96.6% for DeepSeek-R1 and 24.2% for Mixtral 8x22B.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $2 / $6 for Mixtral 8x22B.
- DeepSeek-R1 accepts more context: 164K tokens versus 64K.
- Mixtral 8x22B has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | Mixtral 8x22B | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 42.3 | 27.1 |
| Released | 2025-01-20 | 2024-04-17 |
| Weights | Proprietary | Open |
| Context window | 164K | 64K |
| Max output | 64K | 64K |
| Input $ / M tokens | $0.50 | $2 |
| Output $ / M tokens | $2.15 | $6 |
| Results tracked | 52 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), Mixtral 8x22B: 24.2 (#329)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| WeirdML | 41.6% | 3.2% |
| LMArena Coding | 1427 | 1166 |
| Aider Polyglot | 71.4% | — |
| SciCode | 35.7% | — |
| BigCodeBench Instruct | — | 40.6% |
| LiveBench Coding | 66.7% | — |
| BigCodeBench Complete | — | 50.2% |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
| HumanEval+ | — | 72% |
| MBPP+ | — | 64.3% |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), Mixtral 8x22B: 23.1 (#127)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| Cybench | — | 7.5% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning Mixtral 8x22B leads
DeepSeek-R1: 18.6 (#278), Mixtral 8x22B: 19.9 (#248)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| LMArena Hard Prompts | 1416 | 1150 |
| Epoch Capabilities Index | 141.29 | 122.03 |
| ForecastBench | 60 | 56.3 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| Kagi LLM Benchmark | 69.4% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| LiveBench Reasoning | 83.2% | — |
| DTBench | — | 55.1% |
| LiveBench Data Analysis | 69.8% | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), Mixtral 8x22B: 22.9 (#275)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| Omni-MATH | 42.4% | 16.3% |
| LMArena Math | 1400 | 1184 |
| MATH Level 5 | 96.6% | 24.2% |
| OTIS Mock AIME 2024-2025 | 66.4% | — |
| LiveBench Math | 80.7% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), Mixtral 8x22B: 15.1 (#293)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| GPQA Diamond | 76.3% | 34.1% |
| MMLU-Pro | 79.3% | 46% |
| GPQA (HELM) | 66.6% | 33.4% |
| LMArena Expert | 1394 | 1113 |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| MMLU | — | 77.8% |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), Mixtral 8x22B: 32.8 (#255)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| LMArena Non-English | 1412 | 1128 |
| LMArena Chinese | 1442 | 1116 |
| LMArena French | 1417 | 1166 |
| LMArena German | 1404 | 1141 |
| LMArena Japanese | 1391 | 1037 |
| LMArena Korean | 1360 | 1057 |
| LMArena Russian | 1423 | 1158 |
| LMArena Spanish | 1411 | 1151 |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.0 (#143), Mixtral 8x22B: 57.7 (#266)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| IFEval | 78.4% | 72.4% |
| LMArena Instruction Following | 1382 | 1147 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), Mixtral 8x22B: 34.7 (#247)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| LMArena Longer Query | 1391 | 1144 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), Mixtral 8x22B: 36.9 (#262)
| Benchmark | DeepSeek-R1 | Mixtral 8x22B |
|---|---|---|
| LMArena Text | 1428 | 1162 |
| LMArena Creative Writing | 1405 | 1141 |
| WildBench | 82.8% | 71.1% |
| LMArena Multi-Turn | 1405 | 1130 |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than Mixtral 8x22B?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 27.1 on the Noometry Index.
Which is cheaper, DeepSeek-R1 or Mixtral 8x22B?
DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; Mixtral 8x22B lists at $2 and $6.
Is DeepSeek-R1 or Mixtral 8x22B better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 versus 24.2 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 64K.
How many benchmarks do DeepSeek-R1 and Mixtral 8x22B share?
27 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and Mixtral 8x22B has 34.