Tags: AI, LLMs, Agents, Coding, Anthropic, Claude, tokenomics
So far a couple of hours with Antrhopic's new Fable model in Claude Code feels like it consistently does what Opus does on a good day. Claude Code with Fable more consistently has The Magic ™️ that it had in November and December of last year. It does look like it'll be notably more expensive than Opus, and it does seem to run longer without asking for clarification (for better or worse), but the rate at which it just makes software is pretty impressive.
Once again interesting to see performance curves in the wild. Opus 4.8 at Max reasoning is worse than at Xhigh and more expensive...
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。