Cursor Adds Fable 5.2, Tops CursorBench

Fable 5.2 looks like a meaningful coding upgrade, but its headline advantage depends on paying for higher effort and trusting a vendor-run benchmark.

• –Default CursorBench performance rises from 49.2% for Fable 5.1 to 56.9% for Fable 5.2
• –Max effort reaches 60%, making it the current top model on CursorBench
• –Lower token pricing improves the economics of long-running coding agents and multi-file refactors
• –High, extra-high, and max effort can increase latency and token usage, so developers should benchmark their own repositories
• –Privacy Mode users may need to approve Fable 5.2’s data-retention policy before enabling it