Meta's Muse Spark 1.1 APEX-Agents Score Jumps to 41.9% After Filter Tuning
RylanSchaeffer · x · 2026-07-22
Mercor released an updated score for Meta's Muse Spark 1.1 on the APEX-Agents benchmark.
Previously, the model scored 0 on roughly 10% of long-horizon professional tasks (in banking, law, and consulting) due to false positives in content moderation. After collaborating with Meta to tune the safety filters on the public checkpoint, its Pass@1 score increased from 37.1% to 41.9%. This places the model at #2 overall, just behind Claude Fable 5 (43.3%).
More from Models
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11
- Terminal Bench v4: GLM-5.3 Leads at 41.9%, Kimi-K3 Underwhelms at 12.6% — Ok_Warning2146 · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11