Anthropic announced the release of Claude Opus 4.8 on May 28, 2026, as an upgraded version of its previous model. The update includes improvements in benchmark performance and enhanced collaboration capabilities. Users can now control the amount of effort Claude applies to tasks, with options for fast mode that runs at 2.5× the speed of previous models at 3× lower cost. The release also introduces dynamic workflows in Claude Code, enabling the model to handle large-scale problems by running hundreds of parallel subagents in a single session. Additionally, the Messages API now supports system entries within the messages array, allowing developers to update instructions mid-task without disrupting the prompt cache. Source: [anthropic](https://www.anthropic.com/news/claude-opus-4-8) Opus 4.8 demonstrates significant improvements in coding, agentic skills, and reasoning compared to its predecessor. Early testers noted the model's enhanced reliability and judgment, particularly in complex tasks. On the Super-Agent benchmark, Opus 4.8 completed all cases end-to-end, outperforming prior Opus models and GPT-5.5 at parity on cost. In the Legal Agent Benchmark, it achieved the highest score recorded, breaking 10% on the all-pass standard. The model also excels in practical knowledge work, scoring 84% on the Online-Mind2Web test, which is a notable improvement over both Opus 4.7 and GPT-5.5. Source: [anthropic](https://www.anthropic.com/news/claude-opus-4-8) The update includes enhancements to tool calling and efficiency, with Opus 4.8 using fewer steps for the same intelligence. It maintains end-to-end task execution and improves upon Opus 4.6 while addressing issues from Opus 4.7. The model also shows better honesty, being around four times less likely than its predecessor to allow flaws in code to pass unremarked. Alignment assessments indicate Opus 4.8 has improved prosocial traits and lower rates of misaligned behavior compared to Opus 4.7. Source: [anthropic](https://www.anthropic.com/news/claude-opus-4-8)