Claude Opus 5.5 vs GPT-5.6 Luna
Claude Opus 5.5 
GPT-5.6 Luna Verdict
- Claude Opus 5.5 (thinking) leads on quality: 70.7 vs 60.8.
- Claude Opus 5.5 (thinking) is stronger in composite, knowledge, long context, multimodal, reasoning.
- GPT-5.6 Luna (xhigh) is stronger in agents & tools, coding, human preference, maths.
- GPT-5.6 Luna (xhigh) is 18× cheaper ($0.450 vs $8.00 per 1M blended).
- GPT-5.6 Luna (xhigh) streams 3.0× faster (143 vs 48 tokens per second).
| Metric | Claude Opus 5.5 (thinking) | GPT-5.6 Luna (xhigh) |
|---|---|---|
| BenchLeader Index | 70.7 | 60.8 |
| Composite score | 95.0 | 71.3 |
| Knowledge score | 85.0 | 58.3 |
| Long context score | 68.5 | 67.0 |
| Multimodal score | 71.9 | 61.4 |
| Reasoning score | 95.0 | 59.5 |
| Agents & tools score | – | 54.1 |
| Coding score | – | 59.6 |
| Human preference score | – | 64.8 |
| Maths score | – | 67.0 |
| Blended price $/M | $8.00 | $0.450 |
| Output speed | 48 tok/s | 143 tok/s |
| Time to first answer | 4.2 s | 40.2 s |
| Context window | 1M | 1.1M |
| SimpleBench | – | 46.8% |
| SciCode | – | 50.0% |
| LMArena Text | – | 1453 |
| LMArena Hard Prompts | – | 1473 |
| LMArena Coding | – | 1498 |
| LMArena WebDev | – | 1519 |
| LMArena Vision | – | 1259 |
| LMArena Agent | – | -0.4 |
| AA Intelligence Index | 57.6 | 34.6 |
| AA-LCR | 84.7% | 81.7% |
| MMMU-Pro | 87.7% | 78.5% |
| AA-Omniscience | 46.4 | -10.8 |
| GPQA Diamond (AA) | – | 89.5% |
| Humanity's Last Exam (AA) | 61.4% | 37.0% |
| SciCode (AA) | 66.9% | 50.5% |
| ARC-AGI-1 | – | 87.7% |
| ARC-AGI-2 | – | 47.6% |
| ARC-AGI-3 | – | 0.0% |
| CritPt | 31.7% | 20.6% |
| GDPval (AA) | 67.3% | 44.2% |
| τ²-Bench Banking (AA) | – | 28.7% |
| LMArena Maths | – | 1477 |
| LMArena Creative Writing | – | 1411 |
| LMArena Instruction Following | – | 1442 |
| LMArena Multi-turn | – | 1458 |
| LMArena Longer Queries | – | 1450 |
| LMArena Document | – | 1462 |
| DeepSWE | – | 56.9% |
| LMCA | – | 48.5% |
| DTBench | – | 89.1% |
| CursorBench | – | 33.0% |
| Terminal-Bench 4.0 (AA) | 59.6% | 3.5% |
| Terminal-Bench 2.1 (AA) | – | 77.9% |
| AutomationBench | 69.5% | – |
| GDP.pdf | 26.2% | – |
| Harvey LAB | 91.2% | – |
| AA-Omniscience: accuracy | 66.2% | 42.5% |
| AA-Omniscience: non-hallucination | 41.4% | 7.5% |
| AA-Briefcase | 1822 | – |
Data as of 2026-09-22. Best configuration of each model; every score links to its source on the model pages.
Claude Opus 5.5 vs GPT-5.6 Luna: questions
- Is Claude Opus 5.5 better than GPT-5.6 Luna?
- Claude Opus 5.5 (thinking) leads on quality: 70.7 vs 60.8. The BenchLeader Index combines every independent quality benchmark; Claude Opus 5.5 (thinking) is ahead overall as of 2026-09-22, but check the category scores for your use.
- Which is cheaper, Claude Opus 5.5 or GPT-5.6 Luna?
- GPT-5.6 Luna is cheaper: $0.450 against $8.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Opus 5.5 or GPT-5.6 Luna?
- GPT-5.6 Luna streams faster: 143 against 48 output tokens per second.
- Which has the larger context window?
- GPT-5.6 Luna accepts more context: 1.1M against 1M tokens.