LLM
Claude Fable 5 Review
Anthropic · LLM
Claude Fable 5 retook the best-coding pick at 80.3% SWE-Bench Pro on July 1, 2026 after the US government lifted the June 12 export-control order. It's cur
Claude Fable 5 retook the best-coding pick at 80.3% SWE-Bench Pro on July 1, 2026 after the US government lifted the June 12 export-control order. It's currently the highest-scoring coding model of any usable LLM. Purpose-built for real-world software engineering: multi-file edits, deep codebase understanding, and reliable refactoring.
Pros
- 80.3% SWE-Bench Pro — highest of any usable model
- Excellent at multi-file changes
- Understands large codebases
- Rare hallucinations on code
- Great with legacy code
Cons
- More expensive than Sonnet 5
- Overkill for simple autocomplete
- US-gated for a period earlier
Use cases
- Large refactoring projects
- Legacy code modernization
- Complex bug hunting
- Architecture decisions
- Code review
Pricing
paid — from $30/mo. No free tier.