cot-prompting-requires-62b-plus-params
IN premise — entries/2026/06/21/wiki-Large_language_model-chunk-2.md
Created 2026-06-21T09:50:09+00:00
Chain-of-thought prompting only improves performance for models with at least approximately 62 billion parameters; smaller models perform better answering directly.
Summary
This sets a hard boundary for prompt design: if your system runs a model under roughly 62 billion parameters, forcing it to lay out step-by-step reasoning before answering will actually make it less accurate than just letting it answer directly. The practical implication is that prompting strategy must be matched to model size, not applied uniformly across the system.
Dependents
These beliefs depend on this one:
- OUT cot-threshold-validates-emergent-discontinuity — Chain-of-thought prompting's empirically measured threshold of ~62B parameters is a specific documented instance of emergent abilities' discontinuous appearance at scale, validating that reasoning itself is an emergent property rather than a gradually improving one.