Model comparison
Claude Sonnet 4.6 vs DeepSeek V4 Pro
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 50.3 on the Noometry Index.
Last verified . 42 shared benchmarks.
DeepSeekSummary
- They share 42 benchmarks with published results for both. Claude Sonnet 4.6 scores higher in 5 categories and DeepSeek V4 Pro in 4 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 52.9.
- The biggest single-benchmark swing is Chess Puzzles: 13% for Claude Sonnet 4.6 and 47% for DeepSeek V4 Pro.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.6.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| Claude Sonnet 4.6 | DeepSeek V4 Pro | |
|---|---|---|
| Provider | Anthropic | DeepSeek |
| Noometry Index | 50.3 | 54.3 |
| Released | 2026-02-17 | 2026-04-24 |
| Weights | Proprietary | Open |
| Context window | 1M | 1M |
| Max output | 128K | 393K |
| Input $ / M tokens | $3 | $0.66 |
| Output $ / M tokens | $15 | $1.98 |
| Results tracked | 57 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
Claude Sonnet 4.6: 46.3 (#67), DeepSeek V4 Pro: 52.4 (#34)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| SWE-bench Verified | 75.2% | 77.6% |
| FrontierCode | 24.3% | 28.6% |
| LMArena WebDev | 1522 | 1582 |
| SciCode | 46.8% | 51% |
| WeirdML | 66.1% | 66.2% |
| LMArena Coding | 1504 | 1470 |
| ALE-Bench | 1,327 | 1,403 |
| DeepSWE | 29.9% | — |
Agentic & Tool Use Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 39.1 (#28), DeepSeek V4 Pro: 32.8 (#58)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| APEX-Agents | 43% | 47.3% |
| Vending-Bench 2 | 7,204 | 3,285 |
| Terminal-Bench | 53.4% | — |
| OSWorld 2.0 | 9.3% | — |
| DeepResearch Bench | 54.9% | — |
| OSWorld | 72.1% | — |
| ExploitBench | 23.6% | — |
| GBAEval | 48.8% | — |
| GDP.pdf | 18% | — |
| LMArena Search | 1221 | — |
Reasoning DeepSeek V4 Pro leads
Claude Sonnet 4.6: 46.1 (#45), DeepSeek V4 Pro: 56.5 (#24)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| ARC-AGI-2 | 60.4% | 61.3% |
| NYT Connections (extended) | 80.9% | 91.3% |
| ARC-AGI-1 | 86.5% | 90.5% |
| CritPt | 3.1% | 18% |
| Chess Puzzles | 13% | 47% |
| LMArena Hard Prompts | 1484 | 1461 |
| Mystery Game Puzzles | 16% | 43% |
| DTBench | 89.9% | 93.9% |
| LMCA | 46.5% | 45.5% |
| Epoch Capabilities Index | 152.24 | 155.31 |
| ForecastBench | 62 | 56.1 |
| Kagi LLM Benchmark | — | 53.5% |
| Thematic Generalization | 76.3% | — |
| Surface Evolver Bench | — | 40% |
Math DeepSeek V4 Pro leads
Claude Sonnet 4.6: 52.9 (#49), DeepSeek V4 Pro: 64.8 (#30)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 85.8% | 98.6% |
| ProofBench | 45% | 50% |
| LMArena Math | 1462 | 1455 |
| FrontierMath (Tiers 1-3) | — | 64.6% |
| FrontierMath Tier 4 | — | 26.8% |
| MathArena Final-Answer Competitions | — | 76.6% |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 8.3% | — |
Knowledge DeepSeek V4 Pro leads
Claude Sonnet 4.6: 51.7 (#65), DeepSeek V4 Pro: 59.5 (#31)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| GPQA Diamond | 87.4% | 91.7% |
| SimpleQA Verified | 35.5% | 52.9% |
| Vectara Hallucination Rate | 10.6% | 8.6% |
| LMArena Expert | 1500 | 1464 |
Multimodal Not comparable
Claude Sonnet 4.6: 38.0 (#68), DeepSeek V4 Pro: —
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| LMArena Vision | 1283 | — |
| Blueprint-Bench 2 | 6.7% | — |
| LMArena Document | 1482 | — |
Multilingual Too close to call
Claude Sonnet 4.6: 54.4 (#41), DeepSeek V4 Pro: 54.4 (#45)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| LMArena Non-English | 1440 | 1439 |
| LMArena Chinese | 1491 | 1486 |
| LMArena French | 1465 | 1472 |
| LMArena German | 1428 | 1458 |
| LMArena Japanese | 1420 | 1445 |
| LMArena Korean | 1411 | 1447 |
| LMArena Russian | 1440 | 1453 |
| LMArena Spanish | 1464 | 1458 |
Instruction Following Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 77.4 (#25), DeepSeek V4 Pro: 76.1 (#47)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| LMArena Instruction Following | 1475 | 1448 |
Long Context Too close to call
Claude Sonnet 4.6: 45.3 (#44), DeepSeek V4 Pro: 45.0 (#51)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| LMArena Longer Query | 1479 | 1458 |
| CL-bench Life | — | 13.5% |
Writing & Preference Claude Sonnet 4.6 leads
Claude Sonnet 4.6: 70.2 (#22), DeepSeek V4 Pro: 65.5 (#46)
| Benchmark | Claude Sonnet 4.6 | DeepSeek V4 Pro |
|---|---|---|
| LMArena Text | 1458 | 1451 |
| LMArena Creative Writing | 1435 | 1446 |
| EQ-Bench Creative Writing | 1810 | 1553 |
| EQ-Bench 4 | 1207 | 1166 |
| LMArena Multi-Turn | 1464 | 1467 |
Frequently asked questions
Is Claude Sonnet 4.6 better than DeepSeek V4 Pro?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 50.3 on the Noometry Index.
Which is cheaper, Claude Sonnet 4.6 or DeepSeek V4 Pro?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; Claude Sonnet 4.6 lists at $3 and $15.
Is Claude Sonnet 4.6 or DeepSeek V4 Pro better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 46.3 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do Claude Sonnet 4.6 and DeepSeek V4 Pro share?
42 benchmarks have published results for both models. Claude Sonnet 4.6 has 57 scored results on Noometry and DeepSeek V4 Pro has 48.