Anthropic unveils Claude Opus 4.8 with new effort controls, updated pricing, and enhanced AI capabilities

Claude Opus has been updated to a new version, with the latest release building upon Opus 4.7 through improvements across benchmarks and enhanced collaboration capabilities. Version 4.8 launches with several new features, including greater control over the amount of effort Claude puts into a task.

Claude Code also receives new “dynamic workflows” features that allow it to tackle very large-scale problems. Anthropic says the updated model can now work at 2.5x the speed while being three times cheaper than previous versions.

Anthropic highlights that early testers found Claude Opus 4.8 to be more reliable and sharper in its judgement when performing agentic tasks. The company also says the model has improved in honesty. Anthropic stated that it trains all its AI models to be transparent and avoid making claims they cannot support.

More from Tech

This new variant has specifically been trained to flag uncertainties rather than generating unsupported claims. As a result, Opus 4.8 is reportedly around four times less likely than its predecessors to leave flaws in generated code unremarked.

Detailed assessments conducted before release revealed that Opus 4.8 achieved new highs on measures of prosocial traits, including supporting user autonomy and acting in users’ interests. The tests also reportedly showed that the chances of Opus exhibiting misaligned behaviour were slightly lower than those of Opus 4.7. Here's how the new model compares to other contemporaries.

New updates going live today

Dynamic Workflows

This new feature, currently available in research preview, allows Claude to take on even larger tasks in Claude Code. With the update, Claude can plan work independently, run hundreds of parallel sub-agents within a single session, and verify outputs before reporting results back to users.

Effort Control in Claude AI and Cowork

A new control alongside the model selector now allows users to decide how much effort Claude puts into a response. On higher effort settings, Claude will think more deeply and spend more time generating detailed responses. On lower settings, Claude responds faster while consuming usage limits more slowly.

Anthropic says this gives users direct control over the balance between speed and response quality.

Messages API updates

The Messages API now accepts system entries directly inside the message array. Developers can update Claude’s instructions mid-task without breaking the prompt cache or rerouting updates through a user turn.

This feature can be used to update permissions, token budgets, or environment context dynamically while an agent is running.

Anthropic has also revealed that it is working on developing and releasing models that provide many of the same capabilities as Opus at a lower cost. The company added that a new class of models with even higher intelligence than Opus remains in development.

Highlighting the safeguards being developed for the upcoming Claude Mythos preview, Anthropic stated:

“We’re making swift progress on developing these safeguards and expect to be able to bring Mythos-class models to all our customers in the coming weeks.”

Claude Opus 4.8 is available globally starting today. Pricing for regular usage remains unchanged from Opus 4.7 at $5 per million input tokens and $25 per million output tokens. Fast mode pricing stands at $10 per million input tokens and $50 per million output tokens.

Developers can access Claude Opus 4.8 through the Claude API.