Claude Fable 5.1 Leads SimpleBench at 86.6 Percent

Pseudonymous account scaling01 posted that Fable 5.1 now outscores humans on SimpleBench. The attached image displays a leaderboard table with columns for Rank, Model, Score (AVG@5), and Organization. Claude Fable 5.1 from Anthropic appears in first place as NEW with 86.6 percent, positioned above the Human Baseline entry. The account describes itself as running LisanBench, a custom LLM reasoning benchmark, and posting technical analysis of model scaling and capabilities.