Autonomous Agents Seen As Purer LLMs Than Early ChatGPT Systems
——0——
Gemma 4 and latest Qwen are good enough you can do the experiment locally: minimal tool use, models sort it out.
Counterpoint: effective agents that control their own workflow are "purer" LLMs than early ChatGPT era (which did came with a lot of external controls).
5:48 PM · May 18, 2026 · 1.8K Views
5:49 PM · May 18, 2026 · 671 Views