Meta's Muse Wonder Details Emerge: Active Memory and Focus, Edges Google's Argon on Evals
ChrisGPT · x · 2026-10-13
Meta shared new details on Muse Wonder and third-party evals pit it against Google's Gemini 4 Argon.
Core mechanisms:
- Active memory: the model remembers your habits, preferences and past conversations across days, weeks and months.
- Active focus: trained on real-world failure scenarios to anticipate problems and pursue alternative solutions when stuck — demoed continuously researching rental listings across sources, filtering by personal constraints, and autonomously contacting landlords.
vs Gemini 4 Argon (third-party evals):
- Artificial Analysis Intelligence Index: 56 vs 53
- DeepSWE 1.1: 79% vs 77.9%
- Google still leads AutomationBench (51.3% vs 45.4%) and narrowly on Terminal-Bench 4.0 (57.4% vs 57.1%)
- If Wonder Full prices near Muse Spark ($1.25/M in, $4.25/M out), that's 37.5% cheaper on input and 57.5% on output than Argon
Related event: Meta Unveils Muse Wonder Models with Active Cross-Session Memory(3 posts)→
More from coding & agent
- Google's RRSI adds regularization to recursive self-improvement, +3.3 points OOD — jiqizhixin · 2026-10-13
- "You have a magic coding robot and your big idea is cloning mediocre apps?" — rms80 · 2026-10-13
- Open-source coding agent Aster joins NVIDIA Inception program — saheedniyi_02 · 2026-10-13
- Solo dev's Bezalel hits $100 MRR: one MCP URL gives agents memory, email, money — Rasmic · 2026-10-13
- How Atomic Agent's desktop app is just a plain HTTP client to its OpenAI-compatible server — GapNew4766 · 2026-10-13
- Author bridges Gemini web chat into a native Linux executor, cutting Antigravity CLI turns from 400+ to 0 — Icy-Management-5843 · 2026-10-13