Claude Fable 5.5 (Leaked Checkpoint)
Terminal-Bench 4.0 autonomous shell execution (78.5% target), recursive self-critique, and 1.5M-token repository refactoring.
SWE-bench Verified
84.2%
MMLU-Pro
94.6%
MATH-500
98.4%
Context Window
1,500,000 tokens
| Specification | Value |
|---|---|
| Developer / Lab | Anthropic |
| Parameters Total | Proprietary Frontier Dynamic Reasoner |
| Parameters Active | NeuronCore-v3 Optimized MoE |
| Architecture Type | Dual-Phase Constitutional Reflection Engine + RingAttention 1.5M Buffer |
| License | Proprietary Commercial API (Anthropic Staging) |
| Input Pricing (per 1M) | $3.50 |
| Output Pricing (per 1M) | $17.50 |
Head-to-Head Comparisons with Claude Fable 5.5 (Leaked Checkpoint)
Claude Fable 5.5 (Leaked Checkpoint) vs Gemini 4 Argon →Claude Fable 5.5 (Leaked Checkpoint) vs GPT-6.1 Sol →Claude Fable 5.5 (Leaked Checkpoint) vs Gemini 4 Pro (Leaked Benchmark) →Claude Fable 5.5 (Leaked Checkpoint) vs Space Bunny Alpha (Stealth) →Claude Fable 5.5 (Leaked Checkpoint) vs GPT-6 Sol →Claude Fable 5.5 (Leaked Checkpoint) vs GPT-6 Luna →