GPT-5.4 vs GPT-6 Astra
GPT-5.4 Verdict
- GPT-6 Astra (max) leads on quality: 70.6 vs 63.4.
- GPT-5.4 is stronger in composite, human preference, instruction following, long context, multimodal.
- GPT-6 Astra (max) is stronger in agents & tools, coding, knowledge, reasoning, maths.
- GPT-5.4 is 3.9× cheaper ($5.63 vs $22.00 per 1M blended).
- GPT-5.4 streams 2.4× faster (130 vs 54 tokens per second).
| Metric | GPT-5.4 | GPT-6 Astra (max) |
|---|---|---|
| BenchLeader Index | 63.4 | 70.6 |
| Agents & tools score | 66.1 | 68.2 |
| Coding score | 51.6 | 79.0 |
| Composite score | 78.1 | 73.1 |
| Human preference score | 64.5 | – |
| Instruction following score | 71.7 | – |
| Knowledge score | 66.8 | 79.8 |
| Long context score | 67.6 | – |
| Multimodal score | 64.0 | – |
| Reasoning score | 62.3 | 70.9 |
| Maths score | – | 77.0 |
| Blended price $/M | $5.63 | $22.00 |
| Output speed | 130 tok/s | 54 tok/s |
| Time to first answer | 126.4 s | 328.6 s |
| Context window | 1.1M | 1.1M |
| GPQA Diamond | – | 95.8% |
| FrontierMath Tiers 1–3 | – | 93.7% |
| FrontierMath Tier 4 | – | 97.6% |
| OTIS Mock AIME | – | 100.0% |
| SimpleQA Verified | – | 75.6% |
| Terminal-Bench | 81.8% | 58.2% |
| SciCode | – | 56.5% |
| APEX-Agents | 34.9% | – |
| FrontierCode | – | 53.3% |
| Epoch Capabilities Index | 156.9 | – |
| LMArena Text | 1466 | – |
| LMArena Hard Prompts | 1488 | – |
| LMArena Coding | 1514 | – |
| LMArena WebDev | 1386 | 1796 |
| LMArena Vision | 1293 | – |
| LMArena Agent | – | 12.5 |
| LiveBench | – | 82.2% |
| LiveBench Reasoning | – | 92.7% |
| LiveBench Coding | – | 80.4% |
| LiveBench Agentic Coding | – | 57.3% |
| LiveBench Mathematics | – | 96.8% |
| LiveBench Data Analysis | – | 83.0% |
| LiveBench Language | – | 89.4% |
| AA Intelligence Index | 39.0 | – |
| IFBench | 74.0% | – |
| AA-LCR | 82.0% | – |
| MMMU-Pro | 78.4% | – |
| AA-Omniscience | 5.8 | – |
| Terminal-Bench Hard | 57.6% | – |
| GPQA Diamond (AA) | 92.0% | – |
| Humanity's Last Exam (AA) | 43.7% | – |
| τ²-Bench Telecom (AA) | 87.1% | – |
| Terminal-Bench 2.1 (Vals) | – | 87.3% |
| Vals Index | – | 66.6 |
| HiL-Bench | 9.7% | – |
| EQ-Bench 4 | 1272 | – |
| Kagi LLM Benchmark | 63.8% | – |
| ARC-AGI-1 | – | 97.5% |
| ARC-AGI-2 | – | 95.0% |
| ARC-AGI-3 | – | 62.7% |
Data as of 2026-09-09. Best configuration of each model; every score links to its source on the model pages.