Andon Labs' Muse Wonder debuts at #4 on Vending-Bench 2, a big jump over Muse Spark
jack_w_rae · x · 2026-10-20
Alexandr Wang announced Muse Wonder, a new frontier model claiming SOTA performance on personal and professional agents—"our largest and most capable model yet." Jack Rae amplified the follow-up: Muse Wonder Full ranks #4 on Vending-Bench 2, a huge improvement over Muse Spark.
The attached chart shows money balance over 350+ simulated days: GPT-6 Astra leads at $15.6K, followed by Gemini 4 Argon ($13.3K), Claude Fable 5.1, and Muse Wonder Full ($12K); base Muse Wonder reaches $9.3K.
More from Models
- Multiple leaks point to a new open model tomorrow that fits on a single DGX Spark — ChrisUniverse · 2026-10-20
- OpenAI benchmark: GPT-6.1 Sol hits 97% accuracy migrating COBOL at $0.50 per trial — ChrisGPT · 2026-10-20
- Anthropic rumored to unveil clean 5.5 lineup (Fable, Opus, Sonnet, Haiku) before IPO — signulll · 2026-10-20
- Community speculates Opus 5.5 needs heavy prompt engineering to match mystery game demo — imjustnewatai · 2026-10-20
- Tess-4.5-27B open-sourced: same scores with 25% fewer tokens than its Qwen base — jacek2023 · 2026-10-20
- OpenAI Platform quietly adds a Decisions API playground with Predicate, Choice and Score modes — testingcatalog · 2026-10-20