CursorBench 3.2: Fable 5.1 Max Tops, Grok 4.6 Shows Strong Cost Efficiency
mattyp · x · 2026-09-02
Cursor released the CursorBench 3.2 evaluation results, ranking agents based on ambiguous, multi-file tasks from real Cursor sessions.
Key Highlights:
- Fable 5.1 Max takes the top spot with a score of 73.4, though at a higher cost ($9.64/task).
- Grok 4.6 Extra High performs impressively, ranking 3rd with a score of 70.8 and strong cost efficiency ($2.84/task).
- GPT-5.6 Sol Max ranks 11th (67.2), while Kimi K3 Max is at 28th (60.8).
- Gemini 3.7 Flash High achieves a score of 61.6 at a low cost ($1.20/task).
The data includes scores, cost per task, token consumption, and step counts for each model.
Related event: Cursor Launches CursorBench 3.2: Fable 5.1 Tops Coding Agent Leaderboard(4 posts)→
More from Models
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18
- Polymarket opens betting market on who releases the next Gemini Argon model — Polymarket · 2026-10-18
- Grok matter-of-factly lists which groups user was racist against, exempting Africans and Latin Americans — teortaxesTex · 2026-10-18
- A quick test for benchmaxxed demos: swap in Mirror's Edge and see if it generalizes — teortaxesTex · 2026-10-18
- Did OpenAI distill DeepSeek's gains? Benchmark shows OpenAI's Luna scoring just 38 — teortaxesTex · 2026-10-18