Claude Fable 5.5 (Leaked Checkpoint) vs Qwen 4 Max (Leaked Staging)
A direct, empirical evaluation of Claude Fable 5.5 (Leaked Checkpoint) (Anthropic) against Qwen 4 Max (Leaked Staging) (Alibaba Cloud).
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.
Qwen 4 Max (Leaked Staging)
Developed by Alibaba Cloud · 1.2 Trillion (Sparse MoE) parameters
Repository-scale software engineering outscoring GPT-6 Sol, multilingual enterprise agents across 120 languages, and low-cost 1M-token processing.
| Metric | Claude Fable 5.5 (Leaked Checkpoint) | Qwen 4 Max (Leaked Staging) |
|---|---|---|
| SWE-bench Verified | 84.2% | 78.6% |
| MMLU-Pro | 94.6% | 91.4% |
| MATH-500 | 98.4% | 95.6% |
| Context Window | 1,500,000 tokens | 1,000,000 tokens |
| Input Pricing (per 1M) | $3.50 | $1.20 |
| Output Pricing (per 1M) | $17.50 | $4.80 |