Claude Sonnet 5.5 vs GPT-6 Astra
Claude Sonnet 5.5 
GPT-6 Astra Verdict
- GPT-6 Astra (max) leads on quality: 70.2 vs 67.6.
- Claude Sonnet 5.5 (max) is stronger in agents & tools, long context, maths, reasoning.
- GPT-6 Astra (max) is stronger in coding, composite, knowledge, human preference, multimodal.
- Claude Sonnet 5.5 (max) is 5.0× cheaper ($4.00 vs $20.00 per 1M blended).
- Claude Sonnet 5.5 (max) streams 3.0× faster (141 vs 47 tokens per second).
| Metric | Claude Sonnet 5.5 (max) | GPT-6 Astra (max) |
|---|---|---|
| BenchLeader Index | 67.6 | 70.2 |
| Agents & tools score | 69.5 | 68.8 |
| Coding score | 69.8 | 74.4 |
| Composite score | 72.6 | 80.3 |
| Knowledge score | 65.4 | 79.9 |
| Long context score | 66.6 | 65.6 |
| Maths score | 72.9 | 72.2 |
| Reasoning score | 81.1 | 76.7 |
| Human preference score | – | 67.0 |
| Multimodal score | – | 66.3 |
| Blended price $/M | $4.00 | $20.00 |
| Output speed | 141 tok/s | 47 tok/s |
| Time to first answer | 487.1 s | 383.6 s |
| Context window | 1M | 1.1M |
| GPQA Diamond | 95.6% | 95.8% |
| FrontierMath Tiers 1–3 | 88.8% | 93.7% |
| FrontierMath Tier 4 | 80.5% | 97.6% |
| OTIS Mock AIME | 100.0% | 100.0% |
| SimpleQA Verified | 46.5% | 75.6% |
| Terminal-Bench | 61.8% | 58.2% |
| SciCode | 61.0% | 56.5% |
| WeirdML | – | 93.3% |
| APEX-Agents | 75.5% | – |
| FrontierCode | – | 53.3% |
| ProofBench | 100.0% | – |
| LMArena Text | – | 1475 |
| LMArena Hard Prompts | – | 1500 |
| LMArena Coding | – | 1542 |
| LMArena WebDev | – | 1786 |
| LMArena Vision | – | 1281 |
| LMArena Agent | 12 | 13.1 |
| LiveBench | 75.7% | 82.2% |
| LiveBench Reasoning | 91.6% | 92.7% |
| LiveBench Coding | 91.4% | 80.4% |
| LiveBench Agentic Coding | 56.3% | 57.3% |
| LiveBench Mathematics | 96.1% | 96.8% |
| LiveBench Data Analysis | 59.5% | 83.0% |
| LiveBench Language | 78.0% | 89.4% |
| LiveBench Instruction Following | 56.8% | 75.6% |
| AA Intelligence Index v4.3.2 | 56 | 52.7 |
| AA-LCR | 82.7% | 80.7% |
| MMMU-Pro | – | 86.9% |
| AA-Omniscience | 32.3 | 43.4 |
| GPQA Diamond (AA) | – | 96.1% |
| Humanity's Last Exam (AA) | 55.0% | 54.7% |
| SciCode (AA) | 61.0% | 56.5% |
| IOI | – | 100.0% |
| Terminal-Bench 2.1 (Vals) | – | 87.3% |
| Vals Index | – | 63.1 |
| ARC-AGI-1 | – | 97.5% |
| ARC-AGI-2 | – | 95.0% |
| ARC-AGI-3 | – | 98.6% |
| CritPt | 31.4% | 31.7% |
| GDPval-AA v2.1 | 67.0% | 52.1% |
| τ³-Banking (AA) | – | 41.4% |
| ITBench SRE (AA) | – | 48.6% |
| Analyst Agent (AA) | 57.5% | 51.3% |
| BioMysteryBench | – | 79.3% |
| Code Migration | – | 67.7% |
| CUA-bench | – | 19.2% |
| CyberBench | – | 41.1% |
| Excel Modeling Benchmark | – | 71.7% |
| Finance Agent v2 | – | 53.5% |
| Harvey's Legal Agent Benchmark | – | 5.4% |
| Legal Research Bench | – | 39.4% |
| MedCode | – | 48.5% |
| MedScribe | – | 87.9% |
| MysteryMechanism | – | 53.1% |
| ProgramBench | – | 5.5% |
| SAGE | – | 46.4% |
| SREBench | – | 56.9% |
| Tax Agent Bench | – | 63.3% |
| Terminal-Bench 4.0 (Vals) | – | 59.6% |
| Terminal-Bench Science | – | 62.9% |
| Time Horizon Index: KSP | – | 90.5% |
| Vibe Code Bench 1-100 | – | 27.6% |
| Vibe Code Bench v1.1 | – | 89.6% |
| LMArena Maths | – | 1486 |
| LMArena Creative Writing | – | 1448 |
| LMArena Instruction Following | – | 1469 |
| LMArena Multi-turn | – | 1484 |
| LMArena Longer Queries | – | 1484 |
| LMArena Document | – | 1473 |
| Chess Puzzles | – | 72.0% |
| EBR-bench | – | 76.2% |
| Mystery Game Puzzles | 65.0% | 84.0% |
| BALROG | – | 68.3% |
| DeepSWE v1.1 | – | 73.2% |
| CursorBench | 55.5% | – |
| ALE-Bench | – | 2951.3 |
| GDP.pdf | – | 34.2% |
| FrontierSWE | 61.9% | 65.5% |
| Terminal-Bench 4.0 (AA) | 63.6% | 59.1% |
| Terminal-Bench 2.1 (AA) | – | 88.4% |
| AutomationBench | 71.8% | 68.5% |
| GDP.pdf | 25.8% | 31.0% |
| MLCR | 75.0% | 35.0% |
| Harvey LAB | 2.8% | 8.6% |
| AA-Omniscience: accuracy | 54.0% | 62.6% |
| AA-Omniscience: non-hallucination | 53.0% | 48.7% |
| AA-Briefcase v1.1 | 1823 | 1570 |
Data as of 2026-10-11. Best configuration of each model; every score links to its source on the model pages.
Claude Sonnet 5.5 vs GPT-6 Astra: questions
- Is Claude Sonnet 5.5 better than GPT-6 Astra?
- GPT-6 Astra (max) leads on quality: 70.2 vs 67.6. The BenchLeader Index combines every independent quality benchmark; GPT-6 Astra (max) is ahead overall as of 2026-10-11, but check the category scores for your use.
- Is Claude Sonnet 5.5 better than GPT-6 Astra for coding?
- GPT-6 Astra scores higher in coding (74 vs 70 on the category index, where 50 is average).
- Is Claude Sonnet 5.5 better than GPT-6 Astra for agentic tasks?
- Claude Sonnet 5.5 scores higher in agentic tasks (70 vs 69 on the category index, where 50 is average).
- Which is cheaper, Claude Sonnet 5.5 or GPT-6 Astra?
- Claude Sonnet 5.5 is cheaper: $4.00 against $20.00 per million tokens, blended at three input tokens per output token.
- Which is faster, Claude Sonnet 5.5 or GPT-6 Astra?
- Claude Sonnet 5.5 streams faster: 141 against 47 output tokens per second.
- Which has the larger context window?
- GPT-6 Astra accepts more context: 1.1M against 1M tokens.