Anthropic / MODEL EVIDENCE
4 benchmarks · 6 observations · 4 sources
4 benchmarks · large dot: Claude Sonnet 4-20250514; small dots: closest models where reported
claude-sonnet-4-20250514 (32k thinking) · Aider coding benchmarks · Polyglot · percent correct ↗
56.4% · claude-sonnet-4-20250514 (no thinking) · Aider coding benchmarks · Polyglot · percent correct · Aider code editing benchmark ↗
claude-sonnet-4-20250514 · SEAL Showdown · style-controlled Elo · 2025-09-20 · September 20, 2025 snapshot ↗
1032.9 Elo · claude-sonnet-4-20250514 (Thinking) · SEAL Showdown · style-controlled Elo · 2025-09-20 · September 20, 2025 snapshot · SEAL Showdown Technical Report ↗
claude-sonnet-4-20250514 · FrontierMath-2025-02-28-Public · FrontierMath public · 10 questions · source run score · Epoch AI · public February 2025 sample · historical release ↗
claude-sonnet-4-20250514† · Overall difficulty-adjusted Brier index ↗
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
There are not enough shared scored settings to identify close comparisons reliably.
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
claude-sonnet-4-20250514 · claude-sonnet-4-20250514 (32k thinking) · claude-sonnet-4-20250514 (Thinking) · claude-sonnet-4-20250514 (no thinking) · claude-sonnet-4-20250514†