Anthropic launches Claude Opus 5 with enhanced security safeguards
The latest release from the US-based AI lab addresses cybersecurity concerns raised by the Trump administration while positioning the model as a cost-effective alternative to Fable 5 and OpenAI’s GPT-5.6.
Anthropic has publicly released its latest artificial intelligence model, Claude Opus 5, marking a significant step in the company’s product roadmap following recent tensions with US regulators. The release comes weeks after the company’s latest interactions with the Trump administration regarding cybersecurity standards, a period that also saw the temporary offline status of the Claude Fable 5 model due to government concerns.
The company states that Opus 5 approaches the capabilities of Claude Fable 5 across many domains and demonstrates marked improvements in complex coding tasks. However, Anthropic distinguishes the model’s primary use case, marketing it specifically for enterprise knowledge work and biology applications. While Fable 5 remains the recommended option for ambitious projects and long-horizon AI agent work, Opus 5 is positioned as the ideal everyday model for corporate customers.
A central feature of the Opus 5 release is the implementation of stronger cybersecurity safeguards compared to its predecessor, Opus 4.8. Anthropic describes the new model as the most aligned in the Opus series and the least susceptible to being tricked into misuse. This emphasis on security follows a broader industry shift toward tighter regulatory oversight, with OpenAI’s GPT-5.6 recently undergoing a two-week trial period limited to government-approved entities.
Regarding the regulatory process, Anthropic spokesperson Danielle Ghiglieri confirmed to The Verge that the company continues to work with government partners on independent testing of Opus 5. It remains unclear whether such testing was completed prior to the public rollout, but the company maintains that these collaborations are ongoing. The release also appears to address recent controversies surrounding alleged hidden usage caps and the so-called AI money squeeze, where lab costs are passed on to customers.
Pricing for Opus 5 remains consistent with Opus 4.8, set at $5 per million input tokens and $25 per million output tokens. This rate positions the model at half the cost of Fable 5 and slightly cheaper than OpenAI’s GPT-5.6. The company has also introduced a Fast mode option in research preview, offering higher speeds for double the price, alongside an automatic fallback feature that switches to lower-tier models if safety safeguards block a user’s request.
