Model comparison
Claude Sonnet 4 vs DeepSeek-V2.5 (Sep 2024)
Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 37.0 on the Noometry Index.
Last verified . 18 shared benchmarks.
DeepSeekSummary
- They share 18 benchmarks with published results for both. Claude Sonnet 4 scores higher in 6 categories and DeepSeek-V2.5 (Sep 2024) in 2 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in coding, where Claude Sonnet 4 leads 44.0 to 31.3.
- The biggest single-benchmark swing is Aider Polyglot: 61.3% for Claude Sonnet 4 and 17.8% for DeepSeek-V2.5 (Sep 2024).
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 40.6 | 37.0 |
| Released | 2025-05-22 | 2024-09-06 |
| Weights | Proprietary | Open |
| Context window | 200K | — |
| Max output | 64K | — |
| Input $ / M tokens | $3 | — |
| Output $ / M tokens | $15 | — |
| Results tracked | 58 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Sonnet 4 leads
Claude Sonnet 4: 44.0 (#70), DeepSeek-V2.5 (Sep 2024): 31.3 (#274)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| Aider Polyglot | 61.3% | 17.8% |
| LMArena Coding | 1414 | 1310 |
| SWE-bench Verified (bash only) | 64.9% | — |
| SciCode | 40% | — |
| GSO | 4.9% | — |
| WeirdML | 46.1% | — |
| BigCodeBench Instruct | — | 48.6% |
| BigCodeBench Complete | — | 53.2% |
| ALE-Bench | 655.35 | — |
| HumanEval+ | — | 83.5% |
| MBPP+ | — | 74.1% |
Agentic & Tool Use Not comparable
Claude Sonnet 4: 38.3 (#29), DeepSeek-V2.5 (Sep 2024): —
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
Claude Sonnet 4: 23.0 (#169), DeepSeek-V2.5 (Sep 2024): 24.5 (#145)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Hard Prompts | 1372 | 1289 |
| ARC-AGI-2 | 5.9% | — |
| SimpleBench | 45.5% | — |
| Kagi LLM Benchmark | 73% | — |
| ARC-AGI-1 | 40% | — |
| CritPt | 0.3% | — |
| EnigmaEval | 3.1% | — |
| DTBench | 77.1% | — |
| LMCA | 29% | — |
| Epoch Capabilities Index | 141.69 | — |
| ForecastBench | 60.2 | — |
Math Claude Sonnet 4 leads
Claude Sonnet 4: 41.5 (#71), DeepSeek-V2.5 (Sep 2024): 33.3 (#177)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Math | 1375 | 1289 |
| OTIS Mock AIME 2024-2025 | 71.1% | — |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
| FrontierMath (Feb 2025 set) | 4.1% | — |
| FrontierMath Tier 4 (v1) | 0% | — |
Knowledge Claude Sonnet 4 leads
Claude Sonnet 4: 41.6 (#99), DeepSeek-V2.5 (Sep 2024): 34.2 (#190)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Expert | 1371 | 1266 |
| GPQA Diamond | 79.2% | — |
| Humanity's Last Exam | 7.8% | — |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| Vectara Hallucination Rate | 10.3% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal Not comparable
Claude Sonnet 4: 25.6 (#116), DeepSeek-V2.5 (Sep 2024): —
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Vision | 1191 | — |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual Claude Sonnet 4 leads
Claude Sonnet 4: 46.9 (#151), DeepSeek-V2.5 (Sep 2024): 42.7 (#188)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Non-English | 1333 | 1273 |
| LMArena Chinese | 1351 | 1317 |
| LMArena French | 1364 | 1288 |
| LMArena German | 1331 | 1258 |
| LMArena Japanese | 1301 | 1228 |
| LMArena Korean | 1293 | 1209 |
| LMArena Russian | 1355 | 1289 |
| LMArena Spanish | 1356 | 1248 |
Instruction Following Claude Sonnet 4 leads
Claude Sonnet 4: 71.8 (#141), DeepSeek-V2.5 (Sep 2024): 67.7 (#189)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Instruction Following | 1375 | 1280 |
| IFEval | 84% | — |
Long Context DeepSeek-V2.5 (Sep 2024) leads
Claude Sonnet 4: 33.6 (#254), DeepSeek-V2.5 (Sep 2024): 39.5 (#169)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Longer Query | 1398 | 1301 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference Claude Sonnet 4 leads
Claude Sonnet 4: 57.3 (#127), DeepSeek-V2.5 (Sep 2024): 50.0 (#182)
| Benchmark | Claude Sonnet 4 | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Text | 1351 | 1294 |
| LMArena Creative Writing | 1345 | 1285 |
| LMArena Multi-Turn | 1376 | 1297 |
| Short-Story Creative Writing | 81.4% | — |
| EQ-Bench Creative Writing | 1483 | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than DeepSeek-V2.5 (Sep 2024)?
Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 37.0 on the Noometry Index.
Is Claude Sonnet 4 or DeepSeek-V2.5 (Sep 2024) better for coding?
Claude Sonnet 4 scores higher on coding benchmarks: 44.0 versus 31.3 in the Noometry coding category.
How many benchmarks do Claude Sonnet 4 and DeepSeek-V2.5 (Sep 2024) share?
18 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and DeepSeek-V2.5 (Sep 2024) has 22.