Anthropic releases Claude Opus 4.8 with benchmark gains and new collaboration features

Anthropic has released Claude Opus 4.8, an upgrade to its flagship model that builds on Opus 4.7 with improvements across coding, agentic, reasoning, and knowledge-work benchmarks, Anthropic News announced on May 28, 2026. The company said the model is a more effective collaborator and is available at the same price as its predecessor.

The launch includes several new features. Users on claude.ai can now adjust how much effort Claude expends on a task via a new effort control available on all plans. Claude Code gained a research-preview feature called "dynamic workflows" that lets the AI plan work and run hundreds of parallel subagents in a single session. Additionally, the Messages API now accepts system entries inside the messages array, allowing developers to update instructions mid-task without breaking prompt cache.

Pricing remains unchanged from Opus 4.7 at $5 per million input tokens and $25 per million output tokens, while fast mode—which Anthropic says runs at 2.5× the speed—is now three times cheaper than previous models, costing $10 per million input tokens and $50 per million output tokens. The model defaults to a high-effort setting, with "extra" and "max" options available for difficult tasks.

Anthropic highlighted Opus 4.8's improved honesty, noting that evaluations show it is roughly four times less likely than Opus 4.7 to allow flaws in code to pass unremarked. The company's alignment assessment found higher prosocial traits and substantially lower rates of misaligned behavior compared with Opus 4.7, aligning it closer to Claude Mythos Preview.

Looking ahead, Anthropic said it plans to release cheaper models with similar capabilities and, through Project Glasswing, expects to bring Mythos-class models to all customers in the coming weeks after strengthening cyber safeguards. Claude Opus 4.8 is available immediately via the Claude API under the identifier claude-opus-4-8.