Muse Wonder Full matches Opus 5.5 and Sol 6.1 on VulcanBench but lags on speed
alexandr_wang · x · 2026-10-18
alexandrwang quoted Morgan Linton's completed Muse Wonder Full run on VulcanBench: the model is competitive with Astra, Opus 5.5, and Sol 6.1 overall, but notably slower per task — slower than Opus 5.5, much slower than GPT-6.1 Sol, and far slower than GPT-6 Astra. Speed improvement is in the works.
Key takeaways from Linton's testing: Medium effort is the sweet spot for Muse Wonder Full, and model choice should weigh tokens-per-task and cost-per-task alongside accuracy.
More from Models
- DeepSeek V4.1 Flash as a coding agent: fast at the core, but prone to rabbit-holing on polish — teortaxesTex · 2026-10-18
- Users report Codex usage limits draining abnormally fast, 16% gone in seconds — Angaisb_ · 2026-10-18
- OpenAI's Codex weekly usage reset rolls out, limits now shown as 100% left — ChrisUniverse · 2026-10-18
- TeortaxesTex: DeepSeek best positioned to recover from distillation, but Claude personas will linger — teortaxesTex · 2026-10-18
- Elorian's Glimpse matches Argon on labels but costs 25x more than Gemma 4 31B — arnabdotorg · 2026-10-18
- 5-model blind test: DeepSeek builds Adidas shoe site for $0.23 vs Opus 5.5's $33 — cedric_chee · 2026-10-18