Model comparison
DeepSeek-V3 vs Mistral Small
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 33.4 on the Noometry Index. Mistral Small costs 1.5× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Last verified . 36 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 36 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Mistral Small in 1 category; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3 leads 32.1 to 16.4.
- The biggest single-benchmark swing is LiveBench Coding: 70.9% for DeepSeek-V3 and 36.2% for Mistral Small.
- Mistral Small is cheaper at $0.15 / $0.60 per million input/output tokens, against $0.24 / $0.90 for DeepSeek-V3.
- Mistral Small accepts more context: 262K tokens versus 164K.
Side by side
| DeepSeek-V3 | Mistral Small | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 39.5 | 33.4 |
| Released | 2024-12-26 | 2024-02-26 |
| Weights | Open | Open |
| Context window | 164K | 262K |
| Max output | 164K | 256K |
| Input $ / M tokens | $0.24 | $0.15 |
| Output $ / M tokens | $0.90 | $0.60 |
| Results tracked | 60 | 39 |
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 Small: 34.0 (#247)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| SciCode | 35.8% | 26.5% |
| BigCodeBench Instruct | 50% | 36.1% |
| LiveBench Coding | 70.9% | 36.2% |
| LMArena Coding | 1368 | 1362 |
| BigCodeBench Complete | 62.2% | 46.6% |
| Aider Polyglot | 55.1% | — |
| WeirdML | 36.1% | — |
| ALE-Bench | — | 497.62 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Mistral Small: 28.1 (#93)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 37.1% |
| METR Time Horizons | 49.6% | — |
Reasoning Too close to call
DeepSeek-V3: 20.5 (#236), Mistral Small: 19.8 (#250)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| Kagi LLM Benchmark | 52.3% | 37.8% |
| CritPt | 0% | 0% |
| LiveBench Reasoning | 65.8% | 44.8% |
| LMArena Hard Prompts | 1365 | 1335 |
| DTBench | 64.8% | 70.9% |
| LiveBench Data Analysis | 60.9% | 53.7% |
| LMCA | 15.5% | 20.6% |
| LiveBench | 66.9% | 44% |
| SimpleBench | 27.2% | — |
| BIG-Bench Hard | 87.5% | — |
| Epoch Capabilities Index | 135.94 | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Mistral Small: 16.4 (#293)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 5.8% |
| LiveBench Math | 73.5% | 39.9% |
| LMArena Math | 1373 | 1341 |
| MATH Level 5 | 75.5% | 46.8% |
| Omni-MATH | 40.3% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Mistral Small: 31.0 (#222)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| GPQA Diamond | 67.6% | 47.5% |
| Vectara Hallucination Rate | 6.1% | 5.1% |
| LMArena Expert | 1351 | 1291 |
| MMLU | 87.2% | 68.7% |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| TriviaQA | 82.9% | — |
Multimodal Not comparable
DeepSeek-V3: —, Mistral Small: 33.5 (#96)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| LMArena Vision | — | 1142 |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Mistral Small: 45.5 (#169)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| LMArena Non-English | 1358 | 1315 |
| LMArena Chinese | 1391 | 1340 |
| LMArena French | 1385 | 1337 |
| LMArena German | 1374 | 1340 |
| LMArena Japanese | 1333 | 1275 |
| LMArena Korean | 1319 | 1259 |
| LMArena Russian | 1373 | 1324 |
| LMArena Spanish | 1358 | 1346 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Mistral Small: 66.4 (#209)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 63.7% |
| LMArena Instruction Following | 1345 | 1310 |
| IFEval | 83.2% | — |
Long Context Mistral Small leads
DeepSeek-V3: 34.0 (#253), Mistral Small: 40.4 (#156)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| LMArena Longer Query | 1352 | 1327 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Mistral Small: 52.5 (#171)
| Benchmark | DeepSeek-V3 | Mistral Small |
|---|---|---|
| LMArena Text | 1375 | 1338 |
| LMArena Creative Writing | 1364 | 1305 |
| LMArena Multi-Turn | 1389 | 1344 |
| LiveBench Language | 49.1% | 30.5% |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
Frequently asked questions
Is DeepSeek-V3 better than Mistral Small?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 33.4 on the Noometry Index. Mistral Small costs 1.5× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3 or Mistral Small?
Mistral Small is cheaper. It lists at $0.15 per million input tokens and $0.60 per million output tokens; DeepSeek-V3 lists at $0.24 and $0.90.
Is DeepSeek-V3 or Mistral Small better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 34.0 in the Noometry coding category.
Which has the bigger context window?
Mistral Small does, with 262K tokens against 164K.
How many benchmarks do DeepSeek-V3 and Mistral Small share?
36 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Mistral Small has 39.