Model comparison
Claude Opus 4 vs DeepSeek-V3.2-Exp
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 42.9 on the Noometry Index.
Last verified . 34 shared benchmarks.
DeepSeekSummary
- They share 34 benchmarks with published results for both. Claude Opus 4 scores higher in 3 categories and DeepSeek-V3.2-Exp in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in long context, where DeepSeek-V3.2-Exp leads 47.6 to 39.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 64.4% for Claude Opus 4 and 87.8% for DeepSeek-V3.2-Exp.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $15 / $75 for Claude Opus 4.
- Claude Opus 4 accepts more context: 200K tokens versus 164K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| Claude Opus 4 | DeepSeek-V3.2-Exp | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 42.9 | 44.9 |
| Released | 2025-05-22 | 2025-09-29 |
| Weights | Proprietary | Open |
| Context window | 200K | 164K |
| Max output | 32K | 66K |
| Input $ / M tokens | $15 | $0.26 |
| Output $ / M tokens | $75 | $0.38 |
| Results tracked | 56 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4 leads
Claude Opus 4: 47.8 (#55), DeepSeek-V3.2-Exp: 46.4 (#61)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| SWE-bench Verified (bash only) | 67.6% | 70% |
| Aider Polyglot | 72% | 74.2% |
| WeirdML | 43.7% | 39.5% |
| LMArena Coding | 1442 | 1454 |
| SWE-bench Verified | 70.7% | — |
| LMArena WebDev | — | 1362 |
| SWE-bench Multilingual | — | 59% |
| SciCode | — | 38.9% |
| GSO | 6.9% | — |
| AlgoTune | 1.33 | — |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
Claude Opus 4: 34.5 (#39), DeepSeek-V3.2-Exp: 37.9 (#31)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| Terminal-Bench | — | 39.6% |
| Berkeley Function Calling Leaderboard | — | 56.7% |
| TheAgentCompany | — | 42.9% |
| Cybench | 38% | — |
| DeepResearch Bench | 46.8% | — |
| LMArena Search | 1127 | — |
| METR Time Horizons | 63.9% | — |
| Vending-Bench 2 | — | 1,034 |
Reasoning Claude Opus 4 leads
Claude Opus 4: 27.4 (#104), DeepSeek-V3.2-Exp: 22.0 (#185)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| ARC-AGI-2 | 8.6% | 4% |
| Kagi LLM Benchmark | 74.3% | 52.2% |
| ARC-AGI-1 | 35.7% | 57% |
| CritPt | 0.3% | 2.9% |
| LMArena Hard Prompts | 1399 | 1433 |
| DTBench | 81.6% | 87.7% |
| LMCA | 37.4% | 29.1% |
| Epoch Capabilities Index | 142.67 | 146.28 |
| SimpleBench | 58.8% | — |
| NYT Connections (extended) | — | 36.7% |
| Chess Puzzles | — | 14% |
| EnigmaEval | 5.6% | — |
| Thematic Generalization | — | 65% |
| ForecastBench | 61.1 | — |
Math Too close to call
Claude Opus 4: 40.2 (#80), DeepSeek-V3.2-Exp: 40.8 (#75)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 64.4% | 87.8% |
| LMArena Math | 1391 | 1435 |
| FrontierMath (Feb 2025 set) | 4.5% | 22.1% |
| FrontierMath Tier 4 (v1) | 4.2% | 2.1% |
| MathArena Final-Answer Competitions | — | 57.7% |
| ProofBench | — | 8% |
| Omni-MATH | 61.6% | — |
| MATH Level 5 | 85% | — |
Knowledge DeepSeek-V3.2-Exp leads
Claude Opus 4: 43.9 (#82), DeepSeek-V3.2-Exp: 51.6 (#61)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| GPQA Diamond | 76.3% | 83.4% |
| Vectara Hallucination Rate | 12% | 5.3% |
| LMArena Expert | 1384 | 1436 |
| Humanity's Last Exam | 10.7% | — |
| MMLU-Pro | 87.5% | — |
| Confabulations | 15.9% | — |
| GPQA (HELM) | 70.8% | — |
Multimodal Not comparable
Claude Opus 4: 30.9 (#101), DeepSeek-V3.2-Exp: —
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Vision | 1192 | — |
| GeoBench | 49% | — |
| VPCT | 38% | — |
Multilingual DeepSeek-V3.2-Exp leads
Claude Opus 4: 49.0 (#133), DeepSeek-V3.2-Exp: 52.4 (#85)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Non-English | 1362 | 1409 |
| LMArena Chinese | 1386 | 1461 |
| LMArena French | 1370 | 1435 |
| LMArena German | 1391 | 1440 |
| LMArena Japanese | 1332 | 1377 |
| LMArena Korean | 1318 | 1372 |
| LMArena Russian | 1392 | 1423 |
| LMArena Spanish | 1388 | 1441 |
Instruction Following Claude Opus 4 leads
Claude Opus 4: 77.3 (#26), DeepSeek-V3.2-Exp: 74.7 (#89)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Instruction Following | 1406 | 1413 |
| IFEval | 91.8% | — |
Long Context DeepSeek-V3.2-Exp leads
Claude Opus 4: 39.5 (#167), DeepSeek-V3.2-Exp: 47.6 (#15)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| Fiction.LiveBench | 61.1% | 83.3% |
| LMArena Longer Query | 1422 | 1428 |
| CL-bench | — | 13.2% |
| CL-bench Life | — | 9.5% |
Writing & Preference DeepSeek-V3.2-Exp leads
Claude Opus 4: 61.4 (#85), DeepSeek-V3.2-Exp: 62.5 (#73)
| Benchmark | Claude Opus 4 | DeepSeek-V3.2-Exp |
|---|---|---|
| LMArena Text | 1377 | 1425 |
| LMArena Creative Writing | 1386 | 1403 |
| EQ-Bench Creative Writing | 1580 | 1515 |
| LMArena Multi-Turn | 1396 | 1427 |
| Short-Story Creative Writing | 83.6% | — |
| WildBench | 85.2% | — |
Frequently asked questions
Is Claude Opus 4 better than DeepSeek-V3.2-Exp?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.9 to 42.9 on the Noometry Index.
Which is cheaper, Claude Opus 4 or DeepSeek-V3.2-Exp?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; Claude Opus 4 lists at $15 and $75.
Is Claude Opus 4 or DeepSeek-V3.2-Exp better for coding?
Claude Opus 4 scores higher on coding benchmarks: 47.8 versus 46.4 in the Noometry coding category.
Which has the bigger context window?
Claude Opus 4 does, with 200K tokens against 164K.
How many benchmarks do Claude Opus 4 and DeepSeek-V3.2-Exp share?
34 benchmarks have published results for both models. Claude Opus 4 has 56 scored results on Noometry and DeepSeek-V3.2-Exp has 48.