Claude Opus 4.8 vs GPT-5.4
Claude Opus 4.8 
GPT-5.4 Verdict
- Claude Opus 4.8 leads on quality: 64.7 vs 63.4.
- Claude Opus 4.8 is stronger in coding, composite, human preference, multimodal, reasoning.
- GPT-5.4 is stronger in agents & tools, instruction following, knowledge, long context.
- GPT-5.4 is 1.8× cheaper ($5.63 vs $10.00 per 1M blended).
- GPT-5.4 streams 2.3× faster (130 vs 58 tokens per second).
| Metric | Claude Opus 4.8 | GPT-5.4 |
|---|---|---|
| BenchLeader Index | 64.7 | 63.4 |
| Agents & tools score | 65.0 | 66.1 |
| Coding score | 65.7 | 51.6 |
| Composite score | 81.9 | 78.1 |
| Human preference score | 65.6 | 64.5 |
| Instruction following score | 61.6 | 71.7 |
| Knowledge score | 66.3 | 66.8 |
| Long context score | 65.3 | 67.6 |
| Multimodal score | 64.4 | 64.0 |
| Reasoning score | 64.2 | 62.3 |
| Blended price $/M | $10.00 | $5.63 |
| Output speed | 58 tok/s | 130 tok/s |
| Time to first answer | 29.5 s | 126.4 s |
| Context window | 1M | 1.1M |
| Terminal-Bench | – | 81.8% |
| SimpleBench | 64.8% | – |
| Remote Labor Index | 8.3% | – |
| APEX-Agents | – | 34.9% |
| FrontierCode | 46.5% | – |
| GSO-Bench | 47.1% | – |
| Epoch Capabilities Index | 158.3 | 156.9 |
| LMArena Text | 1473 | 1466 |
| LMArena Hard Prompts | 1503 | 1488 |
| LMArena Coding | 1527 | 1514 |
| LMArena WebDev | 1540 | 1386 |
| LMArena Vision | 1289 | 1293 |
| AA Intelligence Index | 42.0 | 39.0 |
| IFBench | 62.2% | 74.0% |
| AA-LCR | 77.7% | 82.0% |
| MMMU-Pro | – | 78.4% |
| AA-Omniscience | 28.8 | 5.8 |
| Terminal-Bench Hard | 58.3% | 57.6% |
| GPQA Diamond (AA) | 92.0% | 92.0% |
| Humanity's Last Exam (AA) | 48.7% | 43.7% |
| SciCode (AA) | 54.4% | – |
| τ²-Bench Telecom (AA) | 94.4% | 87.1% |
| 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% | 9.7% |
| EQ-Bench 4 | 1281 | 1272 |
| Kagi LLM Benchmark | – | 63.8% |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.