{"slug": "manus-2-0-ships-cascade-32-cheaper-agent-framework", "title": "Manus 2.0 Ships Cascade: 32% Cheaper Agent Framework", "summary": "Manus shipped Manus 2.0 the day before OpenAI's DevDay, headlined by Cascade, a rebuilt agent framework the company claims cuts token usage by 23%, speeds up tasks by 28%, and reduces infrastructure costs by 32%. Cascade replaces the monolithic agent design that loads the full capability stack upfront with a lightweight start that pulls in specialized tools only when a task needs them.", "body_md": "While OpenAI was on stage at DevDay, Manus quietly shipped Manus 2.0 the day before. The headliner is Cascade — a rebuilt agent framework that Manus claims cuts token usage by 23%, speeds up tasks by 28%, and slashes infrastructure costs by 32%. The numbers deserve skepticism. The architectural shift underneath them does not. What Cascade Actually Does Most agent frameworks are monolithic: they load the full capability stack upfront and let every task pay the overhead. Cascade flips that. It starts lightweight — minimal resources, minimal cost — and brings in specialized tools only when a task actually needs […]\n\nThe post", "url": "https://wpnews.pro/news/manus-2-0-ships-cascade-32-cheaper-agent-framework", "canonical_source": "https://byteiota.com/manus-2-cascade-agent-framework/", "published_at": "2026-09-30 12:09:03+00:00", "updated_at": "2026-09-30 12:16:36.646291+00:00", "lang": "en", "topics": ["ai-agents", "artificial-intelligence", "ai-tools", "ai-infrastructure"], "entities": ["Manus", "Manus 2.0", "Cascade", "OpenAI", "DevDay"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/manus-2-0-ships-cascade-32-cheaper-agent-framework", "markdown": "https://wpnews.pro/news/manus-2-0-ships-cascade-32-cheaper-agent-framework.md", "text": "https://wpnews.pro/news/manus-2-0-ships-cascade-32-cheaper-agent-framework.txt", "jsonld": "https://wpnews.pro/news/manus-2-0-ships-cascade-32-cheaper-agent-framework.jsonld"}}