Anthropic Launches Claude Opus 4.8 With New Effort Controls

Anthropic has officially released Claude Opus 4.8, the latest iteration of its flagship AI model, which aims to provide more honest and reliable outputs [1]. The update introduces new features designed to improve collaboration, including adjustable effort settings and a research preview of dynamic workflows [1, 3].

Key takeaways

• Claude Opus 4.8 is designed to be more honest and is four times less likely to overlook flaws in its own code compared to previous versions [1, 3].
• Users can now adjust "Effort" settings—Low, Medium, High, or Max—to balance response speed against the depth of the AI's reasoning [1, 3].
• The model's "fast mode" is now three times cheaper than it was for previous versions [1, 2].
• A new "Dynamic workflows" feature in research preview allows the model to manage hundreds of subagents to solve complex, large-scale tasks [1, 3].

Enhanced Reasoning and User Control

Anthropic claims that Opus 4.8 demonstrates improved judgment, noting that the model is more likely to flag uncertainties rather than making unverified claims [1, 2]. Early testers have described the model as a more effective collaborator that asks clarifying questions and pushes back when a user's plan appears unsound [2, 3]. While Anthropic reports that the model shows benchmark improvements across the board, some users have expressed skepticism regarding these internal performance metrics [2].

To manage how the model processes information, Anthropic has introduced an "Effort" selector [1]. Users can choose between Low, Medium, High, and Max settings; higher settings prompt the AI to think longer and provide more detailed responses, while lower settings prioritize speed and token efficiency [1, 3]. The model defaults to "High" effort, which the company describes as the optimal balance for most tasks [3]. Additionally, the "fast mode" for simpler queries has seen a significant price reduction, making it three times cheaper than the fast mode used in previous models [1, 2].

Scaling Complex Tasks

For more intensive projects, the new "Dynamic workflows" feature allows Claude to act as an autonomous system [3]. By running hundreds of subagents in parallel, the model can plan and execute complex tasks, such as large-scale codebase migrations, while verifying its own output before presenting the results to the user [1, 3].

Why it matters

The release of Opus 4.8 highlights a broader industry shift toward models that prioritize reasoning and self-correction over simple, rapid-fire responses [3]. By focusing on "honesty" and the ability to admit uncertainty, Anthropic is attempting to address common criticisms regarding AI hallucinations and the tendency of chatbots to act as "yes-men" [2, 3]. Looking ahead, the company has indicated that an even more powerful class of models, currently referred to as "Mythos," is in development and expected to arrive in the coming weeks [1, 3]. While some organizations are already testing a "Claude Mythos Preview," Anthropic maintains that models of that intelligence level require additional safety safeguards before a wider release [3].