Model comparison
Claude Fable 5 vs GPT-6 Astra
GPT-6 Astra is the stronger model overall, scoring 70.8 to 66.8 on the Noometry Index.
Last verified . 53 shared benchmarks.
Summary
- They share 53 benchmarks with published results for both. Claude Fable 5 scores higher in 5 categories and GPT-6 Astra in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-6 Astra leads 75.3 to 62.2.
- The biggest single-benchmark swing is Furniture Assembly: 35.8% for Claude Fable 5 and 80% for GPT-6 Astra.
- Both cost about the same: $10 input and $50 output per million tokens.
- GPT-6 Astra accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Fable 5 | GPT-6 Astra | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 66.8 | 70.8 |
| Released | 2026-06-07 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $10 | $10 |
| Output $ / M tokens | $50 | $50 |
| Results tracked | 62 | 56 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Astra leads
Claude Fable 5: 70.6 (#4), GPT-6 Astra: 73.7 (#2)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| DeepSWE | 69.9% | 74.1% |
| FrontierCode | 53.5% | 53.3% |
| LMArena WebDev | 1625 | 1786 |
| FrontierSWE | 47% | 65.5% |
| SciCode | 61% | 56.5% |
| GSO | 78.4% | 79.4% |
| WeirdML | 91.9% | 93.6% |
| LMArena Coding | 1519 | 1487 |
| MirrorCode | 63.9% | 46.7% |
| ALE-Bench | 2,041 | 2,951 |
Agentic & Tool Use Claude Fable 5 leads
Claude Fable 5: 54.0 (#2), GPT-6 Astra: 52.9 (#3)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| APEX-Agents | 63.6% | 64.7% |
| Remote Labor Index | 16.1% | 20.8% |
| GDP.pdf | 30% | 34.2% |
| Vending-Bench 2 | 5,680 | 15,515 |
| τ²-bench Banking | 39.7% | — |
| PostTrainBench | 41.8% | — |
| BALROG | — | 68.3% |
| GBAEval | 74.5% | — |
| LMArena Search | 1230 | — |
Reasoning GPT-6 Astra leads
Claude Fable 5: 76.8 (#6), GPT-6 Astra: 85.1 (#1)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| ARC-AGI-2 | 89.2% | 95% |
| NYT Connections (extended) | 92.7% | 98.1% |
| ARC-AGI-1 | 98.5% | 98.5% |
| CritPt | 28.6% | 31.7% |
| Chess Puzzles | 41% | 72% |
| EBR-Bench | 39.5% | 76.2% |
| LMArena Hard Prompts | 1508 | 1462 |
| Mystery Game Puzzles | 52% | 84% |
| DTBench | 98.4% | 97.3% |
| LMCA | 61.1% | 64.4% |
| Bench to the Future 3 | 0.13 | 0.14 |
| Epoch Capabilities Index | 162.06 | 166.45 |
| SimpleBench | 81.9% | — |
| Kagi LLM Benchmark | 91.4% | — |
| EnigmaEval | 39.3% | — |
| Surface Evolver Bench | 95% | — |
Math GPT-6 Astra leads
Claude Fable 5: 88.5 (#5), GPT-6 Astra: 93.5 (#2)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 87% | 93.7% |
| FrontierMath Tier 4 | 90.2% | 97.6% |
| OTIS Mock AIME 2024-2025 | 100% | 100% |
| ProofBench | 95% | 99% |
| LMArena Math | 1519 | 1465 |
| FrontierMath Erdős | 0% | 2.9% |
Knowledge GPT-6 Astra leads
Claude Fable 5: 62.2 (#25), GPT-6 Astra: 75.3 (#1)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| GPQA Diamond | 85.9% | 95.8% |
| SimpleQA Verified | 70.7% | 75.6% |
| LMArena Expert | 1534 | 1483 |
| Humanity's Last Exam | — | 54.8% |
| Vectara Hallucination Rate | — | 8.7% |
Multimodal GPT-6 Astra leads
Claude Fable 5: 45.3 (#17), GPT-6 Astra: 55.0 (#3)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| LMArena Vision | 1324 | 1281 |
| Blueprint-Bench 2 | 38.6% | 49.7% |
| Furniture Assembly | 35.8% | 80% |
| LMArena Document | 1496 | 1468 |
Multilingual Claude Fable 5 leads
Claude Fable 5: 57.3 (#9), GPT-6 Astra: 53.7 (#61)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| LMArena Non-English | 1481 | 1430 |
| LMArena Chinese | 1543 | 1484 |
| LMArena French | 1505 | 1456 |
| LMArena German | 1486 | 1440 |
| LMArena Japanese | 1506 | 1379 |
| LMArena Korean | 1488 | 1426 |
| LMArena Russian | 1504 | 1436 |
| LMArena Spanish | 1498 | 1407 |
Instruction Following Claude Fable 5 leads
Claude Fable 5: 78.6 (#8), GPT-6 Astra: 76.3 (#44)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| LMArena Instruction Following | 1502 | 1450 |
Long Context Claude Fable 5 leads
Claude Fable 5: 46.3 (#23), GPT-6 Astra: 44.5 (#62)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| LMArena Longer Query | 1509 | 1456 |
Writing & Preference Too close to call
Claude Fable 5: 75.9 (#5), GPT-6 Astra: 75.3 (#7)
| Benchmark | Claude Fable 5 | GPT-6 Astra |
|---|---|---|
| LMArena Text | 1491 | 1441 |
| LMArena Creative Writing | 1494 | 1418 |
| EQ-Bench Creative Writing | 1943 | 2173 |
| LMArena Multi-Turn | 1504 | 1448 |
| EQ-Bench 4 | 1340 | — |
Frequently asked questions
Is Claude Fable 5 better than GPT-6 Astra?
GPT-6 Astra is the stronger model overall, scoring 70.8 to 66.8 on the Noometry Index.
Which is cheaper, Claude Fable 5 or GPT-6 Astra?
GPT-6 Astra is cheaper. It lists at $10 per million input tokens and $50 per million output tokens; Claude Fable 5 lists at $10 and $50.
Is Claude Fable 5 or GPT-6 Astra better for coding?
GPT-6 Astra scores higher on coding benchmarks: 73.7 versus 70.6 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 Fable 5 and GPT-6 Astra share?
53 benchmarks have published results for both models. Claude Fable 5 has 62 scored results on Noometry and GPT-6 Astra has 56.