Anthropic today officially released Claude Opus 4.8, a major upgrade to its most powerful large language model. The key highlights include reduced operational costs and the introduction of "Dynamic Workflows".
Key Developments
Claude Opus 4.8 achieved an 88.6% score on SWE-bench Verified, outperforming OpenAI's GPT-5.5 in several programming and logical reasoning benchmarks. Anthropic also introduced "Fast Mode" priced at $10 per million input tokens (a 3x reduction compared to version 4.7). Most notably, the Dynamic Workflows feature allows Claude Code to automatically spin up hundreds of parallel subagents to manage large-scale software projects, handling everything from writing code to running tests and opening pull requests.
Why It Matters
The deep price cut of Fast Mode makes deploying Opus 4.8 into production environments far more economically viable. For software engineers, Claude Code's ability to coordinate hundreds of subagents is a game-changer for maintaining and upgrading complex codebases. Anthropic also emphasized that Opus 4.8 is its most "honest" model to date, significantly reducing code hallucinations.