GPT-5.4 Pro vs GPT-5.5
GPT-5.4 Pro Verdict
- GPT-5.5 leads on quality: 66.6 vs 64.2.
- GPT-5.4 Pro is stronger in multimodal, reasoning.
- GPT-5.5 is stronger in instruction following, knowledge, agents & tools, coding, composite, human preference, long context.
- GPT-5.5 is 6.0× cheaper ($11.25 vs $67.50 per 1M blended).
- GPT-5.5 streams 87.7× faster (88 vs 1 tokens per second).
| Metric | GPT-5.4 Pro | GPT-5.5 |
|---|---|---|
| BenchLeader Index | 64.2 | 66.6 |
| Instruction following score | 67.3 | 73.3 |
| Knowledge score | 73.2 | 73.7 |
| Multimodal score | 69.8 | 64.9 |
| Reasoning score | 75.1 | 69.7 |
| Agents & tools score | – | 68.4 |
| Coding score | – | 57.3 |
| Composite score | – | 77.7 |
| Human preference score | – | 67.6 |
| Long context score | – | 68.8 |
| Blended price $/M | $67.50 | $11.25 |
| Output speed | 1 tok/s | 88 tok/s |
| Time to first answer | 6.7 s | 62.3 s |
| Context window | 1.1M | 1.1M |
| Humanity's Last Exam | 44.3% | – |
| Terminal-Bench | – | 84.7% |
| SimpleBench | 74.1% | 69.0% |
| Remote Labor Index | – | 6.3% |
| APEX-Agents | – | 38.5% |
| FrontierCode | – | 43.0% |
| Epoch Capabilities Index | 158.9 | 159.1 |
| LMArena Text | – | 1477 |
| LMArena Hard Prompts | – | 1498 |
| LMArena Coding | – | 1509 |
| LMArena WebDev | – | 1458 |
| LMArena Vision | – | 1296 |
| LMArena Agent | – | 3 |
| AA Intelligence Index | – | 38.6 |
| IFBench | – | 75.8% |
| AA-LCR | – | 84.3% |
| MMMU-Pro | – | 79.9% |
| AA-Omniscience | – | 20.5 |
| Terminal-Bench Hard | – | 60.6% |
| GPQA Diamond (AA) | – | 93.5% |
| Humanity's Last Exam (AA) | – | 45.8% |
| SciCode (AA) | – | 55.8% |
| τ²-Bench Telecom (AA) | – | 93.9% |
| MultiChallenge | 69.2% | – |
| VISTA | 53.9% | – |
| MultiNRC | 62.3% | – |
| HiL-Bench | – | 39.7% |
| TutorBench | 56.6% | – |
| EQ-Bench 4 | – | 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.