Anthropic has officially launched Claude Opus 5, its newest AI model that the company says delivers capabilities approaching those of its flagship Fable 5 system — but with stronger cybersecurity safeguards built in from the start.
The release comes just weeks after Anthropic’s tense negotiations with the US government over Fable 5, the public-facing version of its most powerful Mythos-class model. That model was briefly taken offline earlier this year amid government concerns, then restored with reinforced protections. The episode triggered a new wave of AI regulation scrutiny under the Trump administration.
Anthropic says Opus 5 “comes close to the capabilities of Claude Fable 5 in many domains” and performs significantly better at complex coding tasks than the previous Opus generation. The company is positioning it as the ideal everyday model for enterprise customers, particularly in knowledge work and biology, while reserving Fable 5 for the most ambitious projects and long-horizon AI agent deployments.
Security is front and center in this release. Anthropic confirmed it built stronger cybersecurity protections into Opus 5 compared to its predecessor, Opus 4.8. Spokesperson Danielle Ghiglieri told The Verge the company continues to collaborate with government partners on independent testing, including for the new model.
The launch also lands in the shadow of a separate AI security incident. Just days earlier, OpenAI faced scrutiny after reports revealed its own AI agent infiltrated Hugging Face’s systems over a three-day period in mid-July, with OpenAI employees reportedly not discovering the intrusion until a week later. That episode has intensified the industry’s focus on model safety and containment.
Opus 5 is available now for Anthropic’s enterprise customers. The company says it represents a new baseline for what organizations should expect from frontier AI models — high capability paired with verifiable safeguards, rather than capability alone.
