Anthropic launches Claude Opus 4.8 with coding, agent gains
Anthropic released Claude Opus 4.8 on May 28, 2026, improving coding accuracy, agent workflows, Effort Control and Dynamic Workflows while keeping API prices unchanged.
Anthropic activated Claude Opus 4.8 on May 28, 2026, upgrading its flagship model across claude.ai, the Claude API under the name claude-opus-4-8, and major cloud platforms. The release followed earlier code leaks and product sightings that pointed to the update.
API pricing remains at $5 per million input tokens and $25 per million output tokens. A new Fast Mode runs about 2.5× faster and is priced at $10 per million input and $50 per million output, a lower price than prior fast options.
Anthropic reported measurable performance gains for Opus 4.8 compared with Opus 4.7. Internal results show the model is four times less likely to miss flaws in code it generates and makes fewer unsupported claims. Alignment assessments recorded higher prosocial trait scores and lower rates of misaligned behavior. Early testers noted increased reliability, sharper judgment and clearer reporting of task status.
A new Effort Control setting lets users choose how much “thinking effort” Claude applies to a request. Options range from Low for faster, lower-rate-limit responses to Max for deeper, slower reasoning; the model defaults to High.
A research preview called Dynamic Workflows in Claude Code enables the model to plan and decompose very large tasks, run many parallel subagents, verify outputs and coordinate large-scale code changes. Anthropic describes support for migrations and refactors across codebases with hundreds of thousands of lines.
Developers received a Messages API update that permits inserting system instructions mid-conversation without invalidating prompt cache. The change is intended to simplify long-running or stateful agent workflows and to make it easier to change instructions during a session.
Anthropic made the model available immediately for enterprises and developers to test on complex agentic and coding workloads. The company plans lower-cost models with similar capabilities and intends to roll out Mythos-class models in coming weeks after completing enhanced cyber safeguards under an initiative called Project Glasswing.
Anthropic posted on its official account that Opus 4.8 was “Available today at the same price.” The release arrives as competition in large AI models continues to intensify, prompting frequent updates to performance, safety and developer features.








