Model comparison
Claude Opus 5 vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 67.8 on the Noometry Index. Claude Opus 5 costs 2.0× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. Claude Opus 5 scores higher in 5 categories and GPT-6 Astra in 5 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-6 Astra leads 75.3 to 66.8.
- The biggest single-benchmark swing is EBR-Bench: 45.7% for Claude Opus 5 and 76.2% for GPT-6 Astra.
- Claude Opus 5 is cheaper at $5 / $25 per million input/output tokens, against $10 / $50 for GPT-6 Astra.
- GPT-6 Astra accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Opus 5 | GPT-6 Astra | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 67.8 | 70.8 |
| Released | 2026-07-24 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $5 | $10 |
| Output $ / M tokens | $25 | $50 |
| Results tracked | 57 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Claude Opus 5: 67.5 (#5), GPT-6 Astra: 73.7 (#2)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| DeepSWE | 73.6% | 74.1% |
| FrontierCode | 53.4% | 53.3% |
| LMArena WebDev | 1691 | 1786 |
| FrontierSWE | 52% | 65.5% |
| SciCode | 56.4% | 56.5% |
| WeirdML | 91.8% | 93.6% |
| LMArena Coding | 1534 | 1487 |
| ALE-Bench | 2,165 | 2,951 |
| CursorBench | 46.6% | — |
| GSO | — | 79.4% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 5: 55.6 (#1), GPT-6 Astra: 52.9 (#3)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| APEX-Agents | 65.8% | 64.7% |
| BALROG | 63.4% | 68.3% |
| GDP.pdf | 24% | 34.2% |
| Vending-Bench 2 | 11,182 | 15,515 |
| OSWorld 2.0 | 31.4% | — |
| Remote Labor Index | — | 20.8% |
| τ²-bench Banking | 48.7% | — |
| PostTrainBench | 35% | — |
| GBAEval | 79.6% | — |
Reasoning GPT-6 Astra leads
Claude Opus 5: 77.2 (#4), GPT-6 Astra: 85.1 (#1)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 90.4% | 95% |
| NYT Connections (extended) | 94.3% | 98.1% |
| ARC-AGI-1 | 97.5% | 98.5% |
| CritPt | 29.1% | 31.7% |
| Chess Puzzles | 42% | 72% |
| EBR-Bench | 45.7% | 76.2% |
| LMArena Hard Prompts | 1526 | 1462 |
| Mystery Game Puzzles | 59% | 84% |
| DTBench | 97.9% | 97.3% |
| LMCA | 64.5% | 64.4% |
| Bench to the Future 3 | 0.12 | 0.14 |
| Epoch Capabilities Index | 162.78 | 166.45 |
| SimpleBench | 80.6% | — |
Math GPT-6 Astra leads
Claude Opus 5: 86.2 (#8), GPT-6 Astra: 93.5 (#2)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.6% | 93.7% |
| FrontierMath Tier 4 | 73.2% | 97.6% |
| OTIS Mock AIME 2024-2025 | 98.9% | 100% |
| ProofBench | 99% | 99% |
| LMArena Math | 1531 | 1465 |
| FrontierMath Erdős | — | 2.9% |
Knowledge GPT-6 Astra leads
Claude Opus 5: 66.8 (#9), GPT-6 Astra: 75.3 (#1)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 93.9% | 95.8% |
| SimpleQA Verified | 59.9% | 75.6% |
| LMArena Expert | 1557 | 1483 |
| Humanity's Last Exam | — | 54.8% |
| Vectara Hallucination Rate | — | 8.7% |
Multimodal GPT-6 Astra leads
Claude Opus 5: 50.8 (#8), GPT-6 Astra: 55.0 (#3)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1319 | 1281 |
| Blueprint-Bench 2 | 30.4% | 49.7% |
| Furniture Assembly | 60.8% | 80% |
| LMArena Document | 1516 | 1468 |
Multilingual Claude Opus 5 leads
Claude Opus 5: 58.8 (#4), GPT-6 Astra: 53.7 (#61)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1501 | 1430 |
| LMArena Chinese | 1574 | 1484 |
| LMArena French | 1519 | 1456 |
| LMArena German | 1524 | 1440 |
| LMArena Japanese | 1516 | 1379 |
| LMArena Korean | 1521 | 1426 |
| LMArena Russian | 1507 | 1436 |
| LMArena Spanish | 1519 | 1407 |
Instruction Following Claude Opus 5 leads
Claude Opus 5: 79.2 (#7), GPT-6 Astra: 76.3 (#44)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1517 | 1450 |
Long Context Claude Opus 5 leads
Claude Opus 5: 46.5 (#21), GPT-6 Astra: 44.5 (#62)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1515 | 1456 |
Writing & Preference Claude Opus 5 leads
Claude Opus 5: 79.2 (#1), GPT-6 Astra: 75.3 (#7)
| Benchmark | Claude Opus 5 | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1507 | 1441 |
| LMArena Creative Writing | 1491 | 1418 |
| EQ-Bench Creative Writing | 2133 | 2173 |
| LMArena Multi-Turn | 1499 | 1448 |
| EQ-Bench 4 | 1385 | — |
Frequently asked questions
Is Claude Opus 5 better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 67.8 on the Noometry Index. Claude Opus 5 costs 2.0× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Which is cheaper, Claude Opus 5 or GPT-6 Astra?
Claude Opus 5 is cheaper. It lists at $5 per million input tokens and $25 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is Claude Opus 5 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 67.5 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Astra does, with 1.05M tokens against 1M.
How many benchmarks do Claude Opus 5 and GPT-6 Astra share?
50 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and GPT-6 Astra has 56.