GPT-6 Astra vs GPT-6 Luna
GPT-6 Astra Verdict
- GPT-6 Astra (max) leads on quality: 71.6 vs 61.2.
- GPT-6 Astra (max) is stronger in agents & tools, coding, composite, human preference, knowledge, maths, multimodal, reasoning.
- GPT-6 Luna (max) is stronger in long context.
- GPT-6 Luna (max) is 100× cheaper ($0.200 vs $20.00 per 1M blended).
- GPT-6 Luna (max) streams 2.7× faster (157 vs 59 tokens per second).
| Metric | GPT-6 Astra (max) | GPT-6 Luna (max) |
|---|---|---|
| BenchLeader Index | 71.6 | 61.2 |
| Agents & tools score | 68.6 | – |
| Coding score | 77.6 | – |
| Composite score | 82.6 | 73.9 |
| Human preference score | 68.2 | – |
| Knowledge score | 81.4 | 63.2 |
| Long context score | 66.1 | 67.5 |
| Maths score | 76.9 | – |
| Multimodal score | 66.8 | 59.2 |
| Reasoning score | 77.7 | 76.1 |
| Blended price $/M | $20.00 | $0.200 |
| Output speed | 59 tok/s | 157 tok/s |
| Time to first answer | 326.5 s | 142.6 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 | 58.2% | – |
| SciCode | 56.5% | – |
| WeirdML | 93.3% | – |
| FrontierCode | 53.3% | – |
| LMArena Text | 1480 | – |
| LMArena Hard Prompts | 1497 | – |
| LMArena Coding | 1543 | – |
| LMArena WebDev | 1800 | – |
| LMArena Vision | 1279 | – |
| LMArena Agent | 11.5 | – |
| 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% | – |
| LiveBench Instruction Following | 75.6% | – |
| AA Intelligence Index | 52.7 | 37.3 |
| AA-LCR | 80.7% | 83.3% |
| MMMU-Pro | 86.9% | 75.5% |
| AA-Omniscience | 43.4 | 0.7 |
| GPQA Diamond (AA) | 96.1% | – |
| Humanity's Last Exam (AA) | 54.7% | 38.5% |
| SciCode (AA) | 56.5% | 54.6% |
| IOI | 100.0% | – |
| Terminal-Bench 2.1 (Vals) | 87.3% | – |
| Vals Index | 66.6 | – |
| ARC-AGI-1 | 97.5% | – |
| ARC-AGI-2 | 95.0% | – |
| ARC-AGI-3 | 98.6% | – |
| CritPt | 31.7% | 19.4% |
| GDPval (AA) | 52.1% | 43.4% |
| τ²-Bench Banking (AA) | 41.4% | – |
| Analyst Agent (AA) | 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) | 57.1% | – |
| Terminal-Bench Science | 65.7% | – |
| Time Horizon Index: KSP | 90.5% | – |
| Vibe Code Bench 1-100 | 27.6% | – |
| Vibe Code Bench v1.1 | 89.6% | – |
| DrugDiscoveryBench | 68.7% | – |
| LMArena Creative Writing | 1461 | – |
| LMArena Instruction Following | 1461 | – |
| LMArena Multi-turn | 1499 | – |
| LMArena Longer Queries | 1481 | – |
| LMArena Document | 1473 | – |
| Chess Puzzles | 72.0% | – |
| EBR-bench | 76.2% | – |
| Mystery Game Puzzles | 84.0% | – |
| BALROG | 68.3% | – |
| DeepSWE | 73.2% | – |
| ALE-Bench | 2951.3 | – |
| FrontierSWE | 65.5% | – |
| Terminal-Bench 4.0 (AA) | 59.1% | 12.6% |
| Terminal-Bench 2.1 (AA) | 88.4% | – |
| AutomationBench | 68.5% | 53.2% |
| GDP.pdf | 31.0% | 20.4% |
| MLCR | 35.0% | 16.1% |
| AA-Omniscience: accuracy | 62.6% | 43.8% |
| AA-Omniscience: non-hallucination | 48.7% | 23.3% |
| AA-Briefcase | 1569 | 1299 |
Data as of 2026-09-23. Best configuration of each model; every score links to its source on the model pages.
GPT-6 Astra vs GPT-6 Luna: questions
- Is GPT-6 Astra better than GPT-6 Luna?
- GPT-6 Astra (max) leads on quality: 71.6 vs 61.2. The BenchLeader Index combines every independent quality benchmark; GPT-6 Astra (max) is ahead overall as of 2026-09-23, but check the category scores for your use.
- Which is cheaper, GPT-6 Astra or GPT-6 Luna?
- GPT-6 Luna is cheaper: $0.200 against $20.00 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-6 Astra or GPT-6 Luna?
- GPT-6 Luna streams faster: 157 against 59 output tokens per second.
- Which has the larger context window?
- Both accept 1.1M tokens of context.