Model comparison
DeepSeek-V3 vs Magistral Medium
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 35.2 on the Noometry Index.
Last verified . 20 shared benchmarks.
DeepSeek
Mistral AISummary
- They share 20 benchmarks with published results for both. DeepSeek-V3 scores higher in 6 categories and Magistral Medium in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V3 leads 20.5 to 8.6.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 52.3% for DeepSeek-V3 and 16.2% for Magistral Medium.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $2 / $5 for Magistral Medium.
- Magistral Medium accepts more context: 262K tokens versus 164K.
Side by side
| DeepSeek-V3 | Magistral Medium | |
|---|---|---|
| Provider | DeepSeek | Mistral AI |
| Noometry Index | 39.5 | 35.2 |
| Released | 2024-12-26 | 2025-03-17 |
| Weights | Open | Open |
| Context window | 164K | 262K |
| Max output | 164K | 16K |
| Input $ / M tokens | $0.24 | $2 |
| Output $ / M tokens | $0.90 | $5 |
| Results tracked | 60 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Magistral Medium: 39.1 (#161)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| SciCode | 35.8% | 39.2% |
| LMArena Coding | 1368 | 1319 |
| Aider Polyglot | 55.1% | — |
| WeirdML | 36.1% | — |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Magistral Medium: —
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.5 (#236), Magistral Medium: 8.6 (#348)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| Kagi LLM Benchmark | 52.3% | 16.2% |
| CritPt | 0% | 0.3% |
| LMArena Hard Prompts | 1365 | 1267 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | 27.2% | — |
| ARC-AGI-1 | — | 6.1% |
| LiveBench Reasoning | 65.8% | — |
| DTBench | 64.8% | — |
| LiveBench Data Analysis | 60.9% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| Epoch Capabilities Index | 135.94 | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math Magistral Medium leads
DeepSeek-V3: 32.1 (#219), Magistral Medium: 35.1 (#189)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Math | 1373 | 1250 |
| OTIS Mock AIME 2024-2025 | 37.8% | — |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Magistral Medium: 33.5 (#202)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Expert | 1351 | 1223 |
| GPQA Diamond | 67.6% | — |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Magistral Medium: 39.6 (#224)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Non-English | 1358 | 1232 |
| LMArena Chinese | 1391 | 1227 |
| LMArena French | 1385 | 1267 |
| LMArena German | 1374 | 1248 |
| LMArena Japanese | 1333 | 1175 |
| LMArena Korean | 1319 | 1125 |
| LMArena Russian | 1373 | 1224 |
| LMArena Spanish | 1358 | 1271 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Magistral Medium: 66.0 (#211)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Instruction Following | 1345 | 1254 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |
Long Context Magistral Medium leads
DeepSeek-V3: 34.0 (#253), Magistral Medium: 39.3 (#183)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Longer Query | 1352 | 1295 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Magistral Medium: 46.3 (#219)
| Benchmark | DeepSeek-V3 | Magistral Medium |
|---|---|---|
| LMArena Text | 1375 | 1255 |
| LMArena Creative Writing | 1364 | 1245 |
| LMArena Multi-Turn | 1389 | 1275 |
| Short-Story Creative Writing | 77% | — |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than Magistral Medium?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 35.2 on the Noometry Index.
Which is cheaper, DeepSeek-V3 or Magistral Medium?
DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; Magistral Medium lists at $2 and $5.
Is DeepSeek-V3 or Magistral Medium better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 39.1 in the Noometry coding category.
Which has the bigger context window?
Magistral Medium does, with 262K tokens against 164K.
How many benchmarks do DeepSeek-V3 and Magistral Medium share?
20 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Magistral Medium has 22.