Model comparison
Claude Sonnet 5 vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 54.6 on the Noometry Index. Claude Sonnet 5 costs 5.0× less per token, which makes it the better buy when GPT-6 Astra's lead doesn't matter for your workload.
Last verified . 44 shared benchmarks.
Summary
- They share 44 benchmarks with published results for both. Claude Sonnet 5 scores higher in 3 categories and GPT-6 Astra in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-6 Astra leads 85.1 to 49.1.
- The biggest single-benchmark swing is FrontierMath Tier 4: 29.3% for Claude Sonnet 5 and 97.6% for GPT-6 Astra.
- Claude Sonnet 5 is cheaper at $2 / $10 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 Sonnet 5 | GPT-6 Astra | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 54.6 | 70.8 |
| Released | 2026-06-29 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $2 | $10 |
| Output $ / M tokens | $10 | $50 |
| Results tracked | 51 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Claude Sonnet 5: 55.5 (#26), GPT-6 Astra: 73.7 (#2)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| DeepSWE | 53.8% | 74.1% |
| FrontierCode | 42.7% | 53.3% |
| LMArena WebDev | 1541 | 1786 |
| SciCode | 54.3% | 56.5% |
| GSO | 37.3% | 79.4% |
| WeirdML | 68.8% | 93.6% |
| LMArena Coding | 1483 | 1487 |
| ALE-Bench | 1,463 | 2,951 |
| CursorBench | 34.1% | — |
| FrontierSWE | — | 65.5% |
| MirrorCode | — | 46.7% |
Agentic & Tool Use GPT-6 Astra leads
Claude Sonnet 5: 42.8 (#18), GPT-6 Astra: 52.9 (#3)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| APEX-Agents | 54.5% | 64.7% |
| Vending-Bench 2 | 6,378 | 15,515 |
| Remote Labor Index | — | 20.8% |
| BALROG | — | 68.3% |
| GBAEval | 65.3% | — |
| GDP.pdf | — | 34.2% |
| LMArena Search | 1194 | — |
Reasoning GPT-6 Astra leads
Claude Sonnet 5: 49.1 (#39), GPT-6 Astra: 85.1 (#1)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| NYT Connections (extended) | 75.1% | 98.1% |
| CritPt | 16.9% | 31.7% |
| Chess Puzzles | 35% | 72% |
| LMArena Hard Prompts | 1461 | 1462 |
| Mystery Game Puzzles | 35% | 84% |
| DTBench | 92.5% | 97.3% |
| LMCA | 50% | 64.4% |
| Bench to the Future 3 | 0.14 | 0.14 |
| Epoch Capabilities Index | 156.21 | 166.45 |
| ARC-AGI-2 | — | 95% |
| SimpleBench | 60.6% | — |
| ARC-AGI-1 | — | 98.5% |
| EBR-Bench | — | 76.2% |
| Surface Evolver Bench | 60% | — |
| ForecastBench | 61.1 | — |
Math GPT-6 Astra leads
Claude Sonnet 5: 66.2 (#27), GPT-6 Astra: 93.5 (#2)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 65.6% | 93.7% |
| FrontierMath Tier 4 | 29.3% | 97.6% |
| OTIS Mock AIME 2024-2025 | 94.7% | 100% |
| ProofBench | 77% | 99% |
| LMArena Math | 1467 | 1465 |
| FrontierMath Erdős | — | 2.9% |
Knowledge GPT-6 Astra leads
Claude Sonnet 5: 55.6 (#47), GPT-6 Astra: 75.3 (#1)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 90.5% | 95.8% |
| SimpleQA Verified | 33.7% | 75.6% |
| LMArena Expert | 1490 | 1483 |
| Humanity's Last Exam | — | 54.8% |
| Vectara Hallucination Rate | — | 8.7% |
Multimodal GPT-6 Astra leads
Claude Sonnet 5: 42.4 (#31), GPT-6 Astra: 55.0 (#3)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1274 | 1281 |
| Blueprint-Bench 2 | 24.9% | 49.7% |
| LMArena Document | 1466 | 1468 |
| Furniture Assembly | — | 80% |
Multilingual Too close to call
Claude Sonnet 5: 53.8 (#55), GPT-6 Astra: 53.7 (#61)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1431 | 1430 |
| LMArena Chinese | 1477 | 1484 |
| LMArena French | 1460 | 1456 |
| LMArena German | 1440 | 1440 |
| LMArena Japanese | 1422 | 1379 |
| LMArena Korean | 1411 | 1426 |
| LMArena Russian | 1451 | 1436 |
| LMArena Spanish | 1437 | 1407 |
Instruction Following Too close to call
Claude Sonnet 5: 76.3 (#41), GPT-6 Astra: 76.3 (#44)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1452 | 1450 |
Long Context Too close to call
Claude Sonnet 5: 44.8 (#55), GPT-6 Astra: 44.5 (#62)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1463 | 1456 |
Writing & Preference GPT-6 Astra leads
Claude Sonnet 5: 69.2 (#25), GPT-6 Astra: 75.3 (#7)
| Benchmark | Claude Sonnet 5 | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1442 | 1441 |
| LMArena Creative Writing | 1416 | 1418 |
| EQ-Bench Creative Writing | 1794 | 2173 |
| LMArena Multi-Turn | 1454 | 1448 |
| EQ-Bench 4 | 1236 | — |
Frequently asked questions
Is Claude Sonnet 5 better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 54.6 on the Noometry Index. Claude Sonnet 5 costs 5.0× 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 Sonnet 5 or GPT-6 Astra?
Claude Sonnet 5 is cheaper. It lists at $2 per million input tokens and $10 per million output tokens; GPT-6 Astra lists at $10 and $50.
Is Claude Sonnet 5 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 55.5 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 Sonnet 5 and GPT-6 Astra share?
44 benchmarks have published results for both models. Claude Sonnet 5 has 51 scored results on Noometry and GPT-6 Astra has 56.