Anthropic Rolls Out Claude Opus 4.8, Introducing Effort Controls And Advanced Agentic Capabilities

AI research company Anthropic announced that it has released Claude Opus 4.8, an updated version of its Claude Opus model that builds on Opus 4.7 with stronger benchmark performance and improved collaboration capabilities. The company said the new model is available immediately at the same price as its predecessor.

The launch is accompanied by several product updates across Anthropic’s platform. On claude.ai, users can now adjust how much effort Claude applies to a task. Claude Code has also gained a new “dynamic workflows” feature designed to handle larger and more complex assignments. In addition, fast mode for Opus 4.8, which operates at 2.5 times the normal speed, is now priced at one-third of the cost of earlier versions.

Anthropic said early testers found Opus 4.8 to be more dependable and more precise in judgment when handling agentic tasks. The company described honesty as one of the model’s key improvements, noting that it is more likely to acknowledge uncertainty and less likely to present unsupported conclusions. Internal evaluations also indicated that Opus 4.8 is roughly four times less likely than Opus 4.7 to leave flaws in generated code unnoticed.

The company also said the model underwent a detailed alignment assessment before release. According to Anthropic’s Alignment team, Opus 4.8 “reaches new highs on our measures of prosocial traits like supporting user autonomy and acting in the user’s best interest.” The assessment found lower levels of misaligned behavior, including deception and cooperation with misuse, compared with Opus 4.7. Full results are included in the Claude Opus 4.8 System Card.

Among the accompanying updates is dynamic workflows, now in research preview, which allows Claude to plan work and coordinate hundreds of parallel subagents within a single session. Anthropic said the system can verify outputs before returning results and can support codebase-scale migrations involving hundreds of thousands of lines of code. The feature is available in Claude Code for Enterprise, Team, and Max plans.

Anthropic Expands Claude with Effort Controls, API Enhancements, and New Performance Tuning Options

Anthropic also introduced effort control in claude.ai and Cowork, giving users the option to determine how much reasoning Claude applies to a response. Higher effort settings are intended to produce stronger answers, while lower settings prioritize speed and use fewer rate limits. The feature is available across all plans.

For developers, the Messages API now accepts system entries within the messages array, allowing instructions to be updated during a task without disrupting the prompt cache or routing changes through a user turn. Anthropic said this can be used to adjust permissions, token budgets, or environment context while an agent is running.

Opus 4.8 defaults to high effort, which Anthropic said offers the best balance between quality and user experience. Users can also choose “extra” or “max” for more demanding work. The company said it is continuing to develop lower-cost models with similar capabilities, while also working on a new class of higher-intelligence systems under Project Glasswing. Claude Opus 4.8 is available now through the Claude API, with pricing unchanged at $5 per million input tokens and $25 per million output tokens, while fast mode costs $10 and $50 respectively.

The post Anthropic Rolls Out Claude Opus 4.8, Introducing Effort Controls And Advanced Agentic Capabilities appeared first on Metaverse Post.