Sunday, February 1, 2026
Agentic Model Trajectory
People are discussing the evolution and increasing agency of AI models like Claude, comparing them to competitors and highlighting their potential for autonomous agent capabilities.
Mmm I don’t know if I agree with this. The impressive part of this is the autonomy and agent capabilities (which we’ve known since people started really pushing agents in Claude Code), not the particular shape it has taken in the fake Reddit . There were early GPT-2 reddit sims.
In isolation, Moltbook really isn't that new. AIs talking to each other in chatrooms is old news. For me, it's the trajectory of more agentic models (GPT-5.2, Opus 4.5, Gemini 3) -> better agent harnesses (Codex, Claude Code) -> agent swarms (Kimi K2.5) -> fully autonomous