Sonnet 4.5
3 mentions across all digests
Sonnet 4.5 is an Anthropic language model that scored 9.9% on PostTrainBench's autonomous LLM fine-tuning task, compared to Opus 4.6's 23.2% six months later.
ImportAI 449: LLMs training other LLMs; 72B distributed training run; computer vision is harder than generative text
Figma's woes compound with Claude Design
Figma's expansion beyond its core designer base (now only 33% of users) backfires as Claude Design and other AI tools capture non-designer segments at lower costs, exploiting Anthropic's structural advantages in inference efficiency and model capability.
Evolutionary Search for Automated Design of Uncertainty Quantification Methods
Evolutionary search using LLMs designs uncertainty quantification methods 6.7% better than hand-crafted baselines, but reveals divergent model strategies—Claude evolves complex estimators while GPT prefers simpler schemes, with Opus 4.6 unexpectedly regressing.