Model comparison
DeepSeek-V3 vs Mistral Large
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 31.9 on the Noometry Index.
Last verified . 49 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 49 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Mistral Large in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3 leads 57.4 to 40.7.
- The biggest single-benchmark swing is LiveBench Math: 73.5% for DeepSeek-V3 and 42.5% for Mistral Large.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $2 / $6 for Mistral Large.
- DeepSeek-V3 accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3 | Mistral Large | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 39.5 | 31.9 |
| Released | 2024-12-26 | 2024-02-26 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 164K | 16K |
| Input $ / M tokens | $0.24 | $2 |
| Output $ / M tokens | $0.90 | $6 |
| Results tracked | 60 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Mistral Large: 34.3 (#240)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| SciCode | 35.8% | 36.2% |
| BigCodeBench Instruct | 50% | 30% |
| LiveBench Coding | 70.9% | 47.1% |
| LMArena Coding | 1368 | 1277 |
| BigCodeBench Complete | 62.2% | 38.3% |
| HumanEval+ | 86.6% | 62.2% |
| MBPP+ | 73% | 59.5% |
| Aider Polyglot | 55.1% | — |
| WeirdML | 36.1% | — |
| ALE-Bench | — | 264.7 |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Mistral Large: 28.6 (#89)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 38.4% |
| METR Time Horizons | 49.6% | — |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.5 (#236), Mistral Large: 15.8 (#310)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| SimpleBench | 27.2% | 22.5% |
| CritPt | 0% | 0% |
| LiveBench Reasoning | 65.8% | 43.5% |
| LMArena Hard Prompts | 1365 | 1257 |
| DTBench | 64.8% | 65.1% |
| LiveBench Data Analysis | 60.9% | 50.1% |
| LMCA | 15.5% | 16.7% |
| Epoch Capabilities Index | 135.94 | 128.52 |
| ForecastBench | 59.1 | 57.1 |
| LiveBench | 66.9% | 48.4% |
| Kagi LLM Benchmark | 52.3% | — |
| BIG-Bench Hard | 87.5% | — |
| HellaSwag | 88.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Mistral Large: 18.2 (#291)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 8.5% |
| Omni-MATH | 40.3% | 28.1% |
| LiveBench Math | 73.5% | 42.5% |
| LMArena Math | 1373 | 1262 |
| MATH Level 5 | 75.5% | 50.3% |
| FrontierMath (Feb 2025 set) | 1.7% | 0.3% |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Mistral Large: 30.1 (#230)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| GPQA Diamond | 67.6% | 51.3% |
| MMLU-Pro | 72.3% | 59.9% |
| Confabulations | 26.1% | 21.4% |
| Vectara Hallucination Rate | 6.1% | 4.5% |
| GPQA (HELM) | 53.8% | 43.5% |
| LMArena Expert | 1351 | 1232 |
| MMLU | 87.2% | 80% |
| ARC (AI2) Challenge | 95.3% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Mistral Large: 40.0 (#219)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| LMArena Non-English | 1358 | 1237 |
| LMArena Chinese | 1391 | 1240 |
| LMArena French | 1385 | 1325 |
| LMArena German | 1374 | 1254 |
| LMArena Japanese | 1333 | 1188 |
| LMArena Korean | 1319 | 1202 |
| LMArena Russian | 1373 | 1257 |
| LMArena Spanish | 1358 | 1268 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Mistral Large: 67.9 (#191)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 67.9% |
| IFEval | 83.2% | 87.7% |
| LMArena Instruction Following | 1345 | 1249 |
Long Context Mistral Large leads
DeepSeek-V3: 34.0 (#253), Mistral Large: 38.3 (#199)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| LMArena Longer Query | 1352 | 1261 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Mistral Large: 40.7 (#242)
| Benchmark | DeepSeek-V3 | Mistral Large |
|---|---|---|
| LMArena Text | 1375 | 1266 |
| LMArena Creative Writing | 1364 | 1243 |
| Short-Story Creative Writing | 77% | 69% |
| EQ-Bench Creative Writing | 1472 | 985 |
| WildBench | 83% | 80.1% |
| LMArena Multi-Turn | 1389 | 1260 |
| LiveBench Language | 49.1% | 39.4% |
Frequently asked questions
Is DeepSeek-V3 better than Mistral Large?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 31.9 on the Noometry Index.
Which is cheaper, DeepSeek-V3 or Mistral Large?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Mistral Large lists at $2 and $6.
Is DeepSeek-V3 or Mistral Large better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 34.3 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3 and Mistral Large share?
49 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Mistral Large has 51.