Anthropic Launches Claude Opus 4.8 With Smarter Judgment and Faster Performance
Anthropic has released Claude Opus 4.8, the latest upgrade to its flagship AI model, bringing sharper judgment, improved honesty, and a suite of new features while keeping the price unchanged from its predecessor.
Claude Opus 4.8 builds on Opus 4.7 with measurable gains across coding, agentic tasks, reasoning, and knowledge work benchmarks. According to Anthropic’s system card, the model is around four times less likely than Opus 4.7 to let flaws in code it has written go unremarked, a notable step forward in reliability for developers depending on autonomous AI workflows.
One of the standout improvements in Claude Opus 4.8 is what Anthropic describes as better honesty. Rather than confidently pushing through uncertainty, the model is more likely to flag what it does not know, ask the right questions before taking action, and push back when a plan does not hold up. Early testers from companies including Cursor, Devin, and Hebbia noted that the model catches its own mistakes more readily and carries context more consistently across long sessions.
Alongside the model itself, Anthropic is rolling out dynamic workflows in Claude Code, currently in research preview for Enterprise, Team, and Max plan users. The feature allows Claude to plan large tasks, spin up hundreds of parallel subagents in a single session, and verify its outputs before reporting back. Anthropic says this makes it possible for Claude Code with Opus 4.8 to handle full codebase migrations across hundreds of thousands of lines of code from start to merge.
Users on claude.ai also gain a new effort control feature, letting them dial up how deeply Claude thinks through a response or dial it down for faster replies that are easier on rate limits. The control is available across all plans. For developers, the Messages API now accepts system entries inside the messages array, making it possible to update Claude’s instructions mid-task without breaking the prompt cache.
Fast mode for Claude Opus 4.8, which runs the model at 2.5 times its standard speed, is now three times cheaper than fast mode was for previous models, a significant cost reduction for teams running high-volume workloads.
Anthropic’s alignment team assessed Opus 4.8 as reaching new highs on prosocial behaviour, with rates of misaligned behaviour substantially lower than Opus 4.7 and comparable to Claude Mythos Preview, the company’s most safety-aligned model to date.
Claude Opus 4.8 is available now via the Claude API using the model string claude-opus-4-8, priced at $5 per million input tokens and $25 per million output tokens for standard usage. Fast mode is priced at $10 per million input tokens and $50 per million output tokens.
