CursorBench 4.0: Fable 5.1 Max Tops at 51.8%, Muse Spark 1.3 Stands Out on Value
EdwardSun0909 · x · 2026-09-11
Cursor updated CursorBench 4.0, evaluating coding agents on ambiguous, multi-file tasks drawn from real Cursor sessions.
- Top: Fable 5.1 Max leads at 51.8% but costs $17.28/task; Opus 5 Max follows the Fable family at 46.6%.
- Muse Spark 1.3 is the value pick: 41.6% at only $2.64/task.
- Grok 4.6 Extra High scores 41.4%; leerob confirms the score is accurate and teases Grok 4.7 coming soon.
- Others: GPT-5.6 Sol Max 41.7%, Gemini 3.8 Flash High 39.6%, Sonnet 5 Max just 34.1%.
The leaderboard also breaks out cost, token usage, and steps per task for cost/quality tradeoffs.
Related event: CursorBench 4.0 Launches with Harder Tasks; Fable 5.1 Max Tops the Board(6 posts)→
More from coding & agent
- Supercharging LLM classification with logprobs: soft classifiers you can threshold-tune — JnBrymn · 2026-10-28
- Codex daily build series day 23: skipping iCloud plugins for a reset — dotey · 2026-10-28
- Cloudflare Access adds fully customizable pages with HTML + Liquid templates — dinasaur_404 · 2026-10-28
- Opus retried every session 20.6 million times during a memory leak hunt — johnlindquist · 2026-10-28
- Open-Source Skill Teaches AI Agents to Design and Produce Videos — songguoxiansen · 2026-10-28
- Not Everything Needs an Agent: Fixed Steps Are Recipes, Save Agents for Judgment — adamse · 2026-10-28