Tests reveal Claude models trained on protein sequences at scale

·2026.11.17 09:00

Key point

Tests show Claude models trained on protein sequences can predict next amino acids in bacterial proteins, a task no other tested model can perform, while Fable 5.2 outperforms open protein language models.

Details

  • Tests confirm the latest Claude models were trained on protein sequences at scale, enabling them to predict the next amino acid in little-studied bacterial proteins.
  • No other model tested can perform this specific prediction task, and the Claude models fail when the protein sequence is shuffled.
  • Fable 5.2 beats every open protein language model tested in these benchmarks.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.