Claude Fable 5.5 (Leaked Checkpoint) vs GPT-6 Astra
A direct, empirical evaluation of Claude Fable 5.5 (Leaked Checkpoint) (Anthropic) against GPT-6 Astra (OpenAI).
Claude Fable 5.5 (Leaked Checkpoint)
Developed by Anthropic · Proprietary Frontier Dynamic Reasoner parameters
Terminal-Bench 4.0 autonomous shell execution (78.5% target), recursive self-critique, and 1.5M-token repository refactoring.
GPT-6 Astra
Developed by OpenAI · Proprietary Frontier parameters
Complex multi-step engineering, formal math proofs, autonomous computer/browser navigation, and CAD synthesis.
| Metric | Claude Fable 5.5 (Leaked Checkpoint) | GPT-6 Astra |
|---|---|---|
| SWE-bench Verified | 84.2% | 79.4% |
| MMLU-Pro | 94.6% | 91.2% |
| MATH-500 | 98.4% | 98.6% |
| Context Window | 1,500,000 tokens | 1,050,000 tokens |
| Input Pricing (per 1M) | $3.50 | $10.00 |
| Output Pricing (per 1M) | $17.50 | $50.00 |