Scale CEO touts assistant benchmark: Muse scores 9.3, beats Instinct 4-1 on real tasks
alexandr_wang · x · 2026-09-09
Scale CEO Alexandr Wang amplified a head-to-head on assistantbenchmark.com where the AI assistant Muse beat Instinct 4-1 (one tie), scoring 9.3 vs 8.6 across 16 equally-scaled real-world tasks.
- Online tasks: Muse scored 8, proactively offering four refundable hotel options when Chicago inventory was tight and asking whether to adjust budget or dates.
- Travel booking: Instinct took the win with a 10 — it logs into the user's own airline/hotel accounts for full supply, handles check-in and boarding passes; Muse relies on Duffel, so supply is thinner.
- Purchasing & recommendations: Muse shone with itemized totals before charging, virtual cards to hide real card numbers, and constraint-aware restaurant picks.
Note the source is Scale's own CEO promoting its benchmark and assistant, but the per-task breakdowns are concrete enough to serve as an agent-capability comparison reference.
Related event: Scale AI Launches Personal Agent Muse, CEO Amplifies Praise(7 posts)→
More from Models
- GrokBot user: flight booking and shopping all blocked, agent can't do anything useful — lucasmeijer · 2026-10-26
- Vending-bench-style eval: Sol 5.6 appears the most aggressive contract-violator — xuanalogue · 2026-10-26
- Fable 5.5 launch rumors persist for third week, but no frontier model drops this month — patience_cave · 2026-10-26
- Abliterated Model Flops in the Wild: It Got Worried About My Mental Health Instead — ryunuck · 2026-10-26
- OpenAI upgrades Decisions API: 400 questions per call, native voice and image support — OpenAIDevs · 2026-10-26
- Fangcun AI, China's first independent frontier AI safety lab, launches with DeepSeek eval report — JacquesThibs · 2026-10-26