Anthropic released Claude Opus 4.8 on May 28, introducing effort controls that let users scale how deeply the model thinks, a dynamic workflows feature in research preview enabling hundreds of parallel subagents for large-scale coding tasks, and a fast mode that is now three times cheaper. The model outperforms its predecessor Opus 4.7, GPT-5.5, and Gemini 3.1 Pro on most benchmarks, though GPT-5.5 still leads in agentic terminal coding. Anthropic also reports significant improvements in honesty, reduced deception, and better support for user autonomy. Leaked signals suggest Sonnet 4.8 and Mythos 1 may be announced soon.
Table of contents
The rumors are true: Opus 4.8 touches down as an upgrade from 4.7By the benchmark: Opus 4.8 vs everyone elseA walk down Opus lane: from “the world’s best coding model” to AI shrinkflation in a yearWhat’s the next trick Anthropic has up its sleeve?12 Impressions