Fetching from the wire…
01
02
03
04
05
06
07
08
Models2026-08-13 · source-backed
The benchmark spans ~16 categories from software engineering and security to kernel work, games, and debugging, at easy/medium/hard, each task programmatically verified. Claude Opus 5 leads the public snapshot at 43.5%, GPT-5.6 Sol at 34.6%, Claude Fable 5 at 34.0%. It carries an explicit training-corpora canary and was assembled through open community contribution under adversarial review, so it hasn't entered training sets yet. Cleanest agentic-coding signal currently available, and the leaderboard ranks cost and tokens alongside resolution rate.
Each link below shares sources, entities, or timing with this story.
Claude Fable built by Anthropic / Shared entities / Earlier coverage
Linked by a graph relationship (Claude Fable built by Anthropic); both cover Bench, Claude Opus, GPT, Terminal; earlier Bench coverage from 2026-07-08.
Cursor supports Claude Opus / Shared entities / Same source domain / Earlier coverage
Linked by a graph relationship (Cursor supports Claude Opus); both cover Bench, GPT, Terminal; reported by the same outlet (tbench.ai).
Claude Fable built by Anthropic / Shared entities / Earlier coverage / Tension
Linked by a graph relationship (Claude Fable built by Anthropic); both cover Bench, Claude Opus, GPT, Terminal; earlier Bench coverage from 2026-04-21.
Claude Fable built by Anthropic / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Claude Fable built by Anthropic); both cover Bench, Claude Opus; overlapping topics (claude, cost).
Claude Fable uses CUDA / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Claude Fable uses CUDA); both cover Bench, Claude Opus; overlapping topics (claude, cost).
Claude Fable built by Anthropic / Shared entities / Earlier coverage / Tension
Linked by a graph relationship (Claude Fable built by Anthropic); both cover Claude Opus, GPT; earlier Claude Opus coverage from 2026-04-20.
Cursor supports Claude Opus / Shared entities / Shared topic / Earlier coverage
Linked by a graph relationship (Cursor supports Claude Opus); both cover Claude Opus, GPT; overlapping topics (claude, each).
GPT competes with Claude / Shared entities / Earlier coverage
Linked by a graph relationship (GPT competes with Claude); both cover Bench, GPT, Terminal; earlier Bench coverage from 2026-07-14.