Claude Fable 5 beats Opus 4.8 in 917 coding tests

View on Reddit

In a benchmark across 917 coding-agent scenarios, Claude Fable 5 outperformed Opus 4.8 by 0.9 points. The test was conducted by a Tessl employee and posted on Reddit.