Claude Sonnet 5.5 vs GPT-5.5
Claude Sonnet 5.5 
GPT-5.5 Verdict
- Claude Sonnet 5.5 (max) leads on quality: 67.6 vs 65.9.
- Claude Sonnet 5.5 (max) is stronger in agents & tools, coding, composite, maths, reasoning.
- GPT-5.5 (xhigh) is stronger in knowledge, long context, instruction following, multimodal.
- Claude Sonnet 5.5 (max) is 2.8× cheaper ($4.00 vs $11.25 per 1M blended).
- Claude Sonnet 5.5 (max) streams 1.6× faster (141 vs 88 tokens per second).
| Metric | Claude Sonnet 5.5 (max) | GPT-5.5 (xhigh) |
|---|---|---|
| BenchLeader Index | 67.6 | 65.9 |
| Agents & tools score | 69.5 | 61.5 |
| Coding score | 69.8 | 61.4 |
| Composite score | 72.6 | 69.2 |
| Knowledge score | 65.4 | 65.5 |
| Long context score | 66.6 | 67.5 |
| Maths score | 72.9 | 69.9 |
| Reasoning score | 81.1 | 75.8 |
| Instruction following score | – | 73.8 |
| Multimodal score | – | 63.1 |
| Blended price $/M | $4.00 | $11.25 |
| Output speed | 141 tok/s | 88 tok/s |
| Time to first answer | 487.1 s | 39.0 s |
| Context window | 1M | 1.1M |
| GPQA Diamond | 95.6% | – |
| FrontierMath Tiers 1–3 | 88.8% | 85.3% |
| FrontierMath Tier 4 | 80.5% | 72.5% |
| OTIS Mock AIME | 100.0% | – |
| SimpleQA Verified | 46.5% | 63.0% |
| Terminal-Bench | 61.8% | – |
| OSWorld-Verified 2.0 | – | 13.0% |
| SciCode | 61.0% | 56.1% |
| WeirdML | – | 84.9% |
| APEX-Agents | 75.5% | – |
| ProofBench | 100.0% | 50.0% |
| GSO-Bench | – | 40.2% |
| LMArena WebDev | – | 1513 |
| LMArena Agent | 12 | 4.4 |
| LiveBench | 75.7% | 80.2% |
| LiveBench Reasoning | 91.6% | 89.7% |
| LiveBench Coding | 91.4% | 82.2% |
| LiveBench Agentic Coding | 56.3% | 54.0% |
| LiveBench Mathematics | 96.1% | 95.9% |
| LiveBench Data Analysis | 59.5% | 81.6% |
| LiveBench Language | 78.0% | 87.4% |
| LiveBench Instruction Following | 56.8% | 70.7% |
| AA Intelligence Index v4.3.2 | 56 | 38.4 |
| IFBench | – | 75.8% |
| AA-LCR | 82.7% | 84.3% |
| MMMU-Pro | – | 79.9% |
| AA-Omniscience | 32.3 | 20.5 |
| Terminal-Bench Hard | – | 60.6% |
| GPQA Diamond (AA) | – | 93.5% |
| Humanity's Last Exam (AA) | 55.0% | 45.8% |
| SciCode (AA) | 61.0% | 55.8% |
| τ²-Bench Telecom (AA) | – | 93.9% |
| LiveCodeBench | – | 85.3% |
| MMLU-Pro | – | 88.1% |
| LegalBench | – | 86.5% |
| CorpFin | – | 68.4% |
| TaxEval | – | 75.0% |
| SWE-bench (Vals) | – | 82.6% |
| GPQA Diamond (Vals) | – | 93.2% |
| AIME 2026 | – | 100.0% |
| HMMT February 2026 | – | 98.5% |
| MathArena Apex | – | 80.2% |
| MCP Atlas | – | 75.3% |
| ARC-AGI-1 | – | 95.0% |
| ARC-AGI-2 | – | 85.0% |
| CritPt | 31.4% | 27.1% |
| GDPval-AA v2.1 | 67.0% | 42.7% |
| τ³-Banking (AA) | – | 39.0% |
| ITBench SRE (AA) | – | 45.8% |
| Analyst Agent (AA) | 57.5% | 50.0% |
| APEX-Agents (AA) | – | 37.7% |
| CaseLaw v2 | – | 66.2% |
| Code Migration | – | 45.2% |
| Excel Modeling Benchmark | – | 64.5% |
| Finance Agent v2 | – | 51.8% |
| Harvey's Legal Agent Benchmark | – | 3.8% |
| Legal Research Bench | – | 40.4% |
| MedCode | – | 49.1% |
| MedScribe | – | 86.9% |
| MMMU-Pro (Vals) | – | 88.3% |
| MortgageTax | – | 68.8% |
| ProgramBench | – | 0.5% |
| Public Benefits Bench | – | 60.9% |
| SAGE | – | 51.5% |
| SkillsBench | – | 62.2% |
| SREBench | – | 3.8% |
| Tax Agent Bench | – | 60.5% |
| Time Horizon Index: KSP | – | 8.3% |
| Vals Multimodal Index | – | 68.1% |
| Vibe Code Bench v1.1 | – | 69.8% |
| FORTRESS | – | 16.3% |
| SWE Atlas: Codebase QnA | – | 45.4% |
| SWE Atlas: Refactoring | – | 44.8% |
| SWE Atlas: Test Writing | – | 42.6% |
| EBR-bench | – | 34.3% |
| Mystery Game Puzzles | 65.0% | 56.0% |
| PostTrainBench | – | 27.2% |
| DeepSWE v1.1 | – | 67.0% |
| LMCA | – | 54.3% |
| DTBench | – | 96.0% |
| CursorBench | 55.5% | – |
| ALE-Bench | – | 1943.0 |
| GDP.pdf | – | 26.0% |
| FrontierSWE | 61.9% | – |
| Terminal-Bench 4.0 (AA) | 63.6% | 14.7% |
| Terminal-Bench 2.1 (AA) | – | 84.3% |
| AutomationBench | 71.8% | – |
| GDP.pdf | 25.8% | – |
| MLCR | 75.0% | – |
| Harvey LAB | 2.8% | – |
| AA-Omniscience: accuracy | 54.0% | 58.0% |
| AA-Omniscience: non-hallucination | 53.0% | 11.0% |
| AA-Briefcase v1.1 | 1823 | – |
Data as of 2026-10-11. Best configuration of each model; every score links to its source on the model pages.
Claude Sonnet 5.5 vs GPT-5.5: questions
- Is Claude Sonnet 5.5 better than GPT-5.5?
- Claude Sonnet 5.5 (max) leads on quality: 67.6 vs 65.9. The BenchLeader Index combines every independent quality benchmark; Claude Sonnet 5.5 (max) is ahead overall as of 2026-10-11, but check the category scores for your use.
- Is Claude Sonnet 5.5 better than GPT-5.5 for coding?
- Claude Sonnet 5.5 scores higher in coding (70 vs 61 on the category index, where 50 is average).
- Is Claude Sonnet 5.5 better than GPT-5.5 for agentic tasks?
- Claude Sonnet 5.5 scores higher in agentic tasks (70 vs 62 on the category index, where 50 is average).
- Which is cheaper, Claude Sonnet 5.5 or GPT-5.5?
- Claude Sonnet 5.5 is cheaper: $4.00 against $11.25 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Sonnet 5.5 or GPT-5.5?
- Claude Sonnet 5.5 streams faster: 141 against 88 output tokens per second.
- Which has the larger context window?
- GPT-5.5 accepts more context: 1.1M against 1M tokens.