Claude Sonnet 5.5 vs GPT-5.6 Luna
Claude Sonnet 5.5 
GPT-5.6 Luna Verdict
- Claude Sonnet 5.5 (max) leads on quality: 67.6 vs 59.6.
- Claude Sonnet 5.5 (max) is stronger in agents & tools, coding, composite, knowledge, long context, maths, reasoning.
- GPT-5.6 Luna (xhigh) is stronger in human preference, multimodal.
- GPT-5.6 Luna (xhigh) is 8.9× cheaper ($0.450 vs $4.00 per 1M blended).
| Metric | Claude Sonnet 5.5 (max) | GPT-5.6 Luna (xhigh) |
|---|---|---|
| BenchLeader Index | 67.6 | 59.6 |
| Agents & tools score | 69.5 | 52.9 |
| Coding score | 69.8 | 57.6 |
| Composite score | 72.6 | 68.7 |
| Knowledge score | 65.4 | 56.8 |
| Long context score | 66.6 | 66.1 |
| Maths score | 72.9 | 66.3 |
| Reasoning score | 81.1 | 58.0 |
| Human preference score | – | 64.2 |
| Multimodal score | – | 60.5 |
| Blended price $/M | $4.00 | $0.450 |
| Output speed | 141 tok/s | 111 tok/s |
| Time to first answer | 487.1 s | 52.9 s |
| Context window | 1M | 1.1M |
| GPQA Diamond | 95.6% | – |
| FrontierMath Tiers 1–3 | 88.8% | – |
| FrontierMath Tier 4 | 80.5% | – |
| OTIS Mock AIME | 100.0% | – |
| SimpleQA Verified | 46.5% | – |
| Terminal-Bench | 61.8% | – |
| SimpleBench | – | 46.8% |
| SciCode | 61.0% | 50.0% |
| APEX-Agents | 75.5% | – |
| ProofBench | 100.0% | – |
| LMArena Text | – | 1452 |
| LMArena Hard Prompts | – | 1474 |
| LMArena Coding | – | 1501 |
| LMArena WebDev | – | 1519 |
| LMArena Vision | – | 1258 |
| LMArena Agent | 12 | -1.5 |
| LiveBench | 75.7% | – |
| LiveBench Reasoning | 91.6% | – |
| LiveBench Coding | 91.4% | – |
| LiveBench Agentic Coding | 56.3% | – |
| LiveBench Mathematics | 96.1% | – |
| LiveBench Data Analysis | 59.5% | – |
| LiveBench Language | 78.0% | – |
| LiveBench Instruction Following | 56.8% | – |
| AA Intelligence Index v4.3.2 | 56 | 34.6 |
| AA-LCR | 82.7% | 81.7% |
| MMMU-Pro | – | 78.5% |
| AA-Omniscience | 32.3 | -10.8 |
| GPQA Diamond (AA) | – | 89.5% |
| Humanity's Last Exam (AA) | 55.0% | 37.0% |
| SciCode (AA) | 61.0% | 50.5% |
| ARC-AGI-1 | – | 87.7% |
| ARC-AGI-2 | – | 47.6% |
| ARC-AGI-3 | – | 0.0% |
| CritPt | 31.4% | 20.6% |
| GDPval-AA v2.1 | 67.0% | 45.0% |
| τ³-Banking (AA) | – | 28.7% |
| Analyst Agent (AA) | 57.5% | – |
| LMArena Maths | – | 1477 |
| LMArena Creative Writing | – | 1410 |
| LMArena Instruction Following | – | 1443 |
| LMArena Multi-turn | – | 1457 |
| LMArena Longer Queries | – | 1452 |
| LMArena Document | – | 1462 |
| Mystery Game Puzzles | 65.0% | – |
| DeepSWE v1.1 | – | 56.9% |
| LMCA | – | 48.5% |
| DTBench | – | 89.1% |
| CursorBench | 55.5% | 33.0% |
| FrontierSWE | 61.9% | – |
| Terminal-Bench 4.0 (AA) | 63.6% | 3.5% |
| Terminal-Bench 2.1 (AA) | – | 77.9% |
| AutomationBench | 71.8% | – |
| GDP.pdf | 25.8% | – |
| MLCR | 75.0% | – |
| Harvey LAB | 2.8% | – |
| AA-Omniscience: accuracy | 54.0% | 42.5% |
| AA-Omniscience: non-hallucination | 53.0% | 7.5% |
| 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-5.6 Luna: questions
- Is Claude Sonnet 5.5 better than GPT-5.6 Luna?
- Claude Sonnet 5.5 (max) leads on quality: 67.6 vs 59.6. 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-5.6 Luna for coding?
- Claude Sonnet 5.5 scores higher in coding (70 vs 58 on the category index, where 50 is average).
- Is Claude Sonnet 5.5 better than GPT-5.6 Luna for agentic tasks?
- Claude Sonnet 5.5 scores higher in agentic tasks (70 vs 53 on the category index, where 50 is average).
- Which is cheaper, Claude Sonnet 5.5 or GPT-5.6 Luna?
- GPT-5.6 Luna is cheaper: $0.450 against $4.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Sonnet 5.5 or GPT-5.6 Luna?
- Claude Sonnet 5.5 streams faster: 141 against 111 output tokens per second.
- Which has the larger context window?
- GPT-5.6 Luna accepts more context: 1.1M against 1M tokens.