Model comparison
Claude Opus 4 vs Claude Sonnet 4.5
Claude Sonnet 4.5 is the stronger model overall, scoring 44.0 to 42.9 on the Noometry Index.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. Claude Opus 4 scores higher in 3 categories and Claude Sonnet 4.5 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Opus 4 leads 40.2 to 31.9.
- The biggest single-benchmark swing is ARC-AGI-1: 35.7% for Claude Opus 4 and 63.7% for Claude Sonnet 4.5.
- Claude Sonnet 4.5 is cheaper at $3 / $15 per million input/output tokens, against $15 / $75 for Claude Opus 4.
- Claude Sonnet 4.5 accepts more context: 1M tokens versus 200K.
Side by side
| Claude Opus 4 | Claude Sonnet 4.5 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Noometry Index | 42.9 | 44.0 |
| Released | 2025-05-22 | 2025-09-29 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1M |
| Max output | 32K | 64K |
| Input $ / M tokens | $15 | $3 |
| Output $ / M tokens | $75 | $15 |
| Results tracked | 56 | 73 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude Opus 4: 47.8 (#55), Claude Sonnet 4.5: 47.8 (#54)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| SWE-bench Verified | 70.7% | 71.3% |
| SWE-bench Verified (bash only) | 67.6% | 71.4% |
| GSO | 6.9% | 14.7% |
| WeirdML | 43.7% | 47.7% |
| LMArena Coding | 1442 | 1489 |
| AlgoTune | 1.33 | 1.52 |
| Aider Polyglot | 72% | — |
| LMArena WebDev | — | 1393 |
| SWE-bench Multilingual | — | 67% |
| SciCode | — | 44.7% |
| ALE-Bench | — | 796.15 |
Agentic & Tool Use Claude Sonnet 4.5 leads
Claude Opus 4: 34.5 (#39), Claude Sonnet 4.5: 38.1 (#30)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| Cybench | 38% | 60% |
| DeepResearch Bench | 46.8% | 52.6% |
| LMArena Search | 1127 | 1159 |
| METR Time Horizons | 63.9% | 67.4% |
| Terminal-Bench | — | 46.5% |
| Berkeley Function Calling Leaderboard | — | 73.2% |
| GDPval | — | 42.5% |
| Remote Labor Index | — | 2.1% |
| τ²-bench Airline | — | 72% |
| τ²-bench Banking | — | 25.3% |
| τ²-bench Retail | — | 72.4% |
| τ²-bench Telecom | — | 84.9% |
| OSWorld | — | 62.9% |
| Vending-Bench 2 | — | 3,839 |
Reasoning Too close to call
Claude Opus 4: 27.4 (#104), Claude Sonnet 4.5: 26.9 (#113)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| ARC-AGI-2 | 8.6% | 13.6% |
| SimpleBench | 58.8% | 54.3% |
| Kagi LLM Benchmark | 74.3% | 57.9% |
| ARC-AGI-1 | 35.7% | 63.7% |
| CritPt | 0.3% | 1.1% |
| EnigmaEval | 5.6% | 6% |
| LMArena Hard Prompts | 1399 | 1462 |
| DTBench | 81.6% | 83.2% |
| LMCA | 37.4% | 38.8% |
| Epoch Capabilities Index | 142.67 | 146.84 |
| ForecastBench | 61.1 | 61.9 |
| NYT Connections (extended) | — | 37.3% |
| Chess Puzzles | — | 12% |
| EBR-Bench | — | 2.4% |
| Mystery Game Puzzles | — | 17% |
Math Claude Opus 4 leads
Claude Opus 4: 40.2 (#80), Claude Sonnet 4.5: 31.9 (#190)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 64.4% | 77.8% |
| Omni-MATH | 61.6% | 55.3% |
| LMArena Math | 1391 | 1448 |
| MATH Level 5 | 85% | 97.7% |
| FrontierMath (Feb 2025 set) | 4.5% | 15.2% |
| FrontierMath Tier 4 (v1) | 4.2% | 4.2% |
| FrontierMath (Tiers 1-3) | — | 23.9% |
| FrontierMath Tier 4 | — | 2.4% |
| ProofBench | — | 19% |
Knowledge Claude Sonnet 4.5 leads
Claude Opus 4: 43.9 (#82), Claude Sonnet 4.5: 48.4 (#70)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| GPQA Diamond | 76.3% | 82.3% |
| Humanity's Last Exam | 10.7% | 13.7% |
| MMLU-Pro | 87.5% | 86.9% |
| Vectara Hallucination Rate | 12% | 12% |
| GPQA (HELM) | 70.8% | 68.6% |
| LMArena Expert | 1384 | 1483 |
| SimpleQA Verified | — | 30.7% |
| Confabulations | 15.9% | — |
Multimodal Claude Sonnet 4.5 leads
Claude Opus 4: 30.9 (#101), Claude Sonnet 4.5: 34.1 (#84)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| VPCT | 38% | 39.8% |
| LMArena Vision | 1192 | — |
| GeoBench | 49% | — |
| LMArena Document | — | 1450 |
Multilingual Claude Sonnet 4.5 leads
Claude Opus 4: 49.0 (#133), Claude Sonnet 4.5: 53.6 (#65)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| LMArena Non-English | 1362 | 1425 |
| LMArena Chinese | 1386 | 1459 |
| LMArena French | 1370 | 1459 |
| LMArena German | 1391 | 1427 |
| LMArena Japanese | 1332 | 1390 |
| LMArena Korean | 1318 | 1403 |
| LMArena Russian | 1392 | 1436 |
| LMArena Spanish | 1388 | 1457 |
Instruction Following Claude Opus 4 leads
Claude Opus 4: 77.3 (#26), Claude Sonnet 4.5: 75.2 (#74)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| IFEval | 91.8% | 85% |
| LMArena Instruction Following | 1406 | 1459 |
Long Context Claude Sonnet 4.5 leads
Claude Opus 4: 39.5 (#167), Claude Sonnet 4.5: 45.2 (#45)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| LMArena Longer Query | 1422 | 1476 |
| Fiction.LiveBench | 61.1% | — |
Writing & Preference Claude Sonnet 4.5 leads
Claude Opus 4: 61.4 (#85), Claude Sonnet 4.5: 66.7 (#33)
| Benchmark | Claude Opus 4 | Claude Sonnet 4.5 |
|---|---|---|
| LMArena Text | 1377 | 1439 |
| LMArena Creative Writing | 1386 | 1442 |
| EQ-Bench Creative Writing | 1580 | 1678 |
| WildBench | 85.2% | 85.4% |
| LMArena Multi-Turn | 1396 | 1465 |
| Short-Story Creative Writing | 83.6% | — |
Frequently asked questions
Is Claude Opus 4 better than Claude Sonnet 4.5?
Claude Sonnet 4.5 is the stronger model overall, scoring 44.0 to 42.9 on the Noometry Index.
Which is cheaper, Claude Opus 4 or Claude Sonnet 4.5?
Claude Sonnet 4.5 is cheaper. It lists at $3 per million input tokens and $15 per million output tokens; Claude Opus 4 lists at $15 and $75.
Is Claude Opus 4 or Claude Sonnet 4.5 better for coding?
They score almost the same on coding (47.8 vs 47.8); test both on your own repository before choosing.
Which has the bigger context window?
Claude Sonnet 4.5 does, with 1M tokens against 200K.
How many benchmarks do Claude Opus 4 and Claude Sonnet 4.5 share?
50 benchmarks have published results for both models. Claude Opus 4 has 56 scored results on Noometry and Claude Sonnet 4.5 has 73.