Model comparison
Claude Opus 5.5 vs Claude Sonnet 4
Claude Opus 5.5 is the stronger model overall, scoring 68.6 to 40.8 on the Noometry Index.
Last verified . 26 shared benchmarks.
Summary
- They share 26 benchmarks with published results for both. Claude Opus 5.5 scores higher in 10 categories and Claude Sonnet 4 in 0 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Claude Opus 5.5 leads 80.2 to 22.9.
- The biggest single-benchmark swing is ARC-AGI-2: 93.3% for Claude Opus 5.5 and 5.9% for Claude Sonnet 4.
- Claude Sonnet 4 is cheaper at $3 / $15 per million input/output tokens, against $4 / $20 for Claude Opus 5.5.
- Claude Opus 5.5 accepts more context: 1M tokens versus 200K.
Side by side
| Claude Opus 5.5 | Claude Sonnet 4 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Noometry Index | 68.6 | 40.8 |
| Released | 2026-09-22 | 2025-05-22 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 200K |
| Max output | 128K | 64K |
| Input $ / M tokens | $4 | $3 |
| Output $ / M tokens | $20 | $15 |
| Results tracked | 44 | 58 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 5.5 leads
Claude Opus 5.5: 71.9 (#3), Claude Sonnet 4: 43.5 (#88)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| SciCode | 66.9% | 40% |
| LMArena Coding | 1547 | 1414 |
| ALE-Bench | 2,147 | 655.35 |
| FrontierCode | 54.6% | — |
| SWE-bench Verified (bash only) | — | 64.9% |
| Aider Polyglot | — | 61.3% |
| CursorBench | 57.8% | — |
| LMArena WebDev | 1813 | — |
| FrontierSWE | 62.3% | — |
| GSO | — | 4.9% |
| WeirdML | — | 46.1% |
| MirrorCode | 77.4% | — |
Agentic & Tool Use Claude Opus 5.5 leads
Claude Opus 5.5: 45.3 (#15), Claude Sonnet 4: 38.5 (#31)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| APEX-Agents | 73.5% | — |
| TheAgentCompany | — | 33.1% |
| Cybench | — | 35% |
| DeepResearch Bench | — | 46.6% |
| OSWorld | — | 43.9% |
| GDP.pdf | 30.6% | — |
| METR Time Horizons | — | 62% |
| Vending-Bench 2 | 9,235 | — |
Reasoning Claude Opus 5.5 leads
Claude Opus 5.5: 80.2 (#3), Claude Sonnet 4: 22.9 (#187)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| ARC-AGI-2 | 93.3% | 5.9% |
| ARC-AGI-1 | 98.5% | 40% |
| CritPt | 31.7% | 0.3% |
| LMArena Hard Prompts | 1535 | 1372 |
| DTBench | 98.9% | 77.1% |
| LMCA | 68.2% | 29% |
| Epoch Capabilities Index | 167.33 | 141.69 |
| SimpleBench | — | 45.5% |
| Kagi LLM Benchmark | — | 73% |
| NYT Connections (extended) | 88.5% | — |
| EnigmaEval | — | 3.1% |
| EBR-Bench | 71.4% | — |
| Mystery Game Puzzles | 71% | — |
| ForecastBench | — | 60.2 |
Math Claude Opus 5.5 leads
Claude Opus 5.5: 91.8 (#3), Claude Sonnet 4: 43.3 (#80)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 100% | 71.1% |
| LMArena Math | 1506 | 1375 |
| FrontierMath (Tiers 1-3) | 91.2% | — |
| FrontierMath Tier 4 | 95% | — |
| ProofBench | 100% | — |
| Omni-MATH | — | 60.2% |
| MATH Level 5 | — | 84.4% |
| FrontierMath (Feb 2025 set) | — | 4.1% |
| FrontierMath Erdős | 2.9% | — |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Claude Opus 5.5 leads
Claude Opus 5.5: 66.4 (#10), Claude Sonnet 4: 41.8 (#108)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| GPQA Diamond | 90.6% | 79.2% |
| LMArena Expert | 1547 | 1372 |
| Humanity's Last Exam | — | 7.8% |
| SimpleQA Verified | 72.2% | — |
| MMLU-Pro | — | 84.3% |
| Confabulations | — | 13.2% |
| Vectara Hallucination Rate | — | 10.3% |
| GPQA (HELM) | — | 70.6% |
Multimodal Claude Opus 5.5 leads
Claude Opus 5.5: 57.8 (#1), Claude Sonnet 4: 26.2 (#121)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| LMArena Vision | 1321 | 1191 |
| GeoBench | — | 37% |
| VPCT | — | 34% |
| Blueprint-Bench 2 | 51.2% | — |
| Furniture Assembly | 83.3% | — |
| MindCube | — | 44.8% |
Multilingual Claude Opus 5.5 leads
Claude Opus 5.5: 59.1 (#2), Claude Sonnet 4: 46.7 (#156)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| LMArena Non-English | 1507 | 1333 |
| LMArena Chinese | 1588 | 1350 |
| LMArena French | 1514 | 1363 |
| LMArena Russian | 1520 | 1355 |
| LMArena Spanish | 1507 | 1357 |
| LMArena German | — | 1331 |
| LMArena Japanese | — | 1302 |
| LMArena Korean | — | 1291 |
Instruction Following Claude Opus 5.5 leads
Claude Opus 5.5: 80.0 (#3), Claude Sonnet 4: 71.7 (#145)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| LMArena Instruction Following | 1537 | 1376 |
| IFEval | — | 84% |
Long Context Claude Opus 5.5 leads
Claude Opus 5.5: 47.1 (#19), Claude Sonnet 4: 33.7 (#259)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| LMArena Longer Query | 1532 | 1398 |
| Fiction.LiveBench | — | 46.9% |
Writing & Preference Claude Opus 5.5 leads
Claude Opus 5.5: 78.2 (#3), Claude Sonnet 4: 57.1 (#132)
| Benchmark | Claude Opus 5.5 | Claude Sonnet 4 |
|---|---|---|
| LMArena Text | 1515 | 1351 |
| LMArena Creative Writing | 1533 | 1345 |
| EQ-Bench Creative Writing | 2050 | 1483 |
| LMArena Multi-Turn | 1499 | 1376 |
| Short-Story Creative Writing | — | 81.4% |
| WildBench | — | 83.8% |
Frequently asked questions
Is Claude Opus 5.5 better than Claude Sonnet 4?
Claude Opus 5.5 is the stronger model overall, scoring 68.6 to 40.8 on the Noometry Index.
Which is cheaper, Claude Opus 5.5 or Claude Sonnet 4?
Claude Sonnet 4 is cheaper. It lists at $3 per million input tokens and $15 per million output tokens; Claude Opus 5.5 lists at $4 and $20.
Is Claude Opus 5.5 or Claude Sonnet 4 better for coding?
Claude Opus 5.5 scores higher on coding benchmarks: 71.9 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
Claude Opus 5.5 does, with 1M tokens against 200K.
How many benchmarks do Claude Opus 5.5 and Claude Sonnet 4 share?
26 benchmarks have published results for both models. Claude Opus 5.5 has 44 scored results on Noometry and Claude Sonnet 4 has 58.