Theo Browne: Please stop using stupid models

Theo argues that a model’s ordinary performance matters more than its best result when an agent runs for a long time. Short tasks hide differences between models; wider tasks create more chances for failure. 04:19 07:24 17:07

He urges teams to learn from long runs and change their prompts and codebases when runs fail. 12:13 12:53

Sources

  1. Please stop using stupid models - YouTube. Theo Browne. Recording; used for model reliability over wider and longer agent runs. Accessed 2026-10-01.

Last changed 2026-10-07 in 4040f67. Source on GitHub. Change log.

Linked from

  1. Agent guide. An Effect stack so pure (aspirational) Kit Langton will blush.

    Theo Browne: Please stop using stupid models — Theo Browne argues that model reliability matters more as agent work gets wider and longer.

  2. Change log. What changed in the files served here, newest first.

    Theo Browne: Please stop using stupid models — edited: feat(learn): teach cards 63 to 82 with plain lines, diagrams and snippets

  3. Floor and ceiling. Theo Browne argues that a model's ordinary reliability matters more than its peak on long agent runs.

    See Theo Browne: Please stop using stupid models.¶

  4. Full agent guide. An Effect stack so pure (aspirational) Kit Langton will blush.
  5. Rat Stack lore | rat-stack. Short, source-grounded notes on the ideas and decisions behind rat-stack.

    Theo Browne: Please stop using stupid models — Theo Browne argues that model reliability matters more as agent work gets wider and longer.

  6. Source change history (log.md). The generated change log as Markdown.

    Theo Browne: Please stop using stupid models — edited: feat(learn): teach cards 63 to 82 with plain lines, diagrams and snippets