Claude Opus 4.8 vs GPT-5.5
Claude Opus 4.8 
GPT-5.5 Verdict
- GPT-5.5 leads on quality: 66.6 vs 64.7.
- Claude Opus 4.8 is stronger in coding, composite.
- GPT-5.5 is stronger in agents & tools, human preference, instruction following, knowledge, long context, multimodal, reasoning.
- They cost about the same ($10.00 per 1M blended).
- GPT-5.5 streams 1.5× faster (88 vs 58 tokens per second).
| Metric | Claude Opus 4.8 | GPT-5.5 |
|---|---|---|
| BenchLeader Index | 64.7 | 66.6 |
| Agents & tools score | 65.0 | 68.4 |
| Coding score | 65.7 | 57.3 |
| Composite score | 81.9 | 77.7 |
| Human preference score | 65.6 | 67.6 |
| Instruction following score | 61.6 | 73.3 |
| Knowledge score | 66.3 | 73.7 |
| Long context score | 65.3 | 68.8 |
| Multimodal score | 64.4 | 64.9 |
| Reasoning score | 64.2 | 69.7 |
| Blended price $/M | $10.00 | $11.25 |
| Output speed | 58 tok/s | 88 tok/s |
| Time to first answer | 29.5 s | 62.3 s |
| Context window | 1M | 1.1M |
| Terminal-Bench | – | 84.7% |
| SimpleBench | 64.8% | 69.0% |
| Remote Labor Index | 8.3% | 6.3% |
| APEX-Agents | – | 38.5% |
| FrontierCode | 46.5% | 43.0% |
| GSO-Bench | 47.1% | – |
| Epoch Capabilities Index | 158.3 | 159.1 |
| LMArena Text | 1473 | 1477 |
| LMArena Hard Prompts | 1503 | 1498 |
| LMArena Coding | 1527 | 1509 |
| LMArena WebDev | 1540 | 1458 |
| LMArena Vision | 1289 | 1296 |
| LMArena Agent | – | 3 |
| AA Intelligence Index | 42.0 | 38.6 |
| IFBench | 62.2% | 75.8% |
| AA-LCR | 77.7% | 84.3% |
| MMMU-Pro | – | 79.9% |
| AA-Omniscience | 28.8 | 20.5 |
| Terminal-Bench Hard | 58.3% | 60.6% |
| GPQA Diamond (AA) | 92.0% | 93.5% |
| Humanity's Last Exam (AA) | 48.7% | 45.8% |
| SciCode (AA) | 54.4% | 55.8% |
| τ²-Bench Telecom (AA) | 94.4% | 93.9% |
| LiveCodeBench | 87.8% | – |
| MMLU-Pro | 89.6% | – |
| LegalBench | 83.6% | – |
| CorpFin | 66.7% | – |
| TaxEval | 75.6% | – |
| Terminal-Bench 2.1 (Vals) | 71.9% | – |
| SWE-bench (Vals) | 88.6% | – |
| GPQA Diamond (Vals) | 92.4% | – |
| Vals Index | 60.9 | – |
| HiL-Bench | 35.3% | 39.7% |
| EQ-Bench 4 | 1281 | 1315 |
| Kagi LLM Benchmark | – | 88.8% |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.