Back
Anthropic launches Claude Opus 5 with adjustable effort settings to cut AI costs

Anthropic launches Claude Opus 5 with adjustable effort settings to cut AI costs

Anthropic priced Claude Opus 5 below Fable 5 and expanded access through API channels, while TrendAI and B.AI outlined enterprise security and large-context deployment plans.

Fact Check
Both core parts of the claim are confirmed by primary sources. Anthropic's official 'Introducing Claude Opus 5' page confirms the July 24, 2026 launch, the adjustable effort settings to optimize for intelligence vs. cost/speed, and the existence of the Cyber Verification Program. The TrendAI PR Newswire release confirms TrendAI's adoption of Claude Opus 5 through the Cyber Verification Program for vulnerability prioritization, assessment, and virtual patching. Fortune corroborates the effort-toggle cost feature, and Anthropic's official X account confirms the launch date.
Summary

Anthropic has released Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, matching Opus 4.8 and undercutting Fable 5 while posting stronger results on several benchmarks. The model adds adjustable effort settings, is available via subscription and API, and is being positioned as the practical default for many paying users. TrendAI said it will adopt Opus 5 through Anthropic's Cyber Verification Program for vulnerability prioritization and virtual patching, while B.AI said its API platform launched Opus 5 with a 1 million-token context window, a 128,000-token single-output limit, and a dual-channel access plan that it said can reduce enterprise costs by up to 40%.

Terms & Concepts
  • virtual patching: A security measure that blocks exploitation of a software flaw before the vendor releases a permanent fix.
  • Zero Data Retention: A setup in which model interactions are not retained, helping organizations meet stricter data-governance requirements.
  • 1 million-token context window: The amount of text and data a model can process in a single prompt session.