Anthropic has released Claude Opus 5, a new AI model presented as a more capable and more carefully guarded step forward for the company’s Opus line. The launch comes after government cybersecurity concerns around Claude Fable 5 and shortly after an OpenAI security incident became a major topic across the tech industry.
The core message is clear: Anthropic wants Opus 5 to look powerful enough for serious work, but safer and more practical than the company’s highest-end model. It is also being priced to keep continuity with its predecessor while sitting below Fable 5.
What Anthropic Says Opus 5 Can Do
Anthropic said Opus 5 “comes close to the capabilities of Claude Fable 5 in many domains” and is much better at complex coding tasks. That places the model close to the company’s public-facing Mythos-class model without making it the new top option for every use case.
The company is especially marketing Claude Opus 5 to enterprise customers. In that positioning, Opus 5 is described as best for knowledge work and biology, and as the ideal everyday model.
Fable 5 still has a separate role in Anthropic’s lineup. The company continues to describe Fable 5 as best for the most ambitious types of work and long-horizon AI agent projects.
That distinction matters because Anthropic is not simply replacing one model with another. It is drawing a line between a model designed for daily enterprise use and a model reserved for more demanding projects.
Cybersecurity Is Central To The Launch
The Opus 5 rollout is taking place in the shadow of concerns over Fable 5. Fable 5 drew the government’s ire, was taken offline for a few weeks along with Mythos 5, and later returned with even stronger cyber safeguards than before.
Those concerns, and the weeks of negotiations between Anthropic and the government that followed, helped begin a new era of AI regulation for the Trump administration. Soon afterward, OpenAI released GPT-5.6 under a roughly two-week trial period, with limited preview access only for government-approved entities.
For Opus 5, Anthropic is emphasizing safeguards from the start. The company said the model has stronger cybersecurity safeguards than Opus 4.8 and called it “the most aligned Opus model and is the least susceptible to being tricked into misuse.”
When asked whether Anthropic ran Opus 5 by the Trump administration before the public rollout, Anthropic spokesperson Danielle Ghiglieri told The Verge, “We continue to work with our government partners to conduct their own independent testing of our models. This includes Opus 5.”
That statement does not describe the full process behind the launch. But it does show that independent government testing is part of Anthropic’s public framing for Opus 5.
How Opus 5 Fits Against Fable 5 And GPT-5.6
Anthropic’s comparison points are doing a lot of work. Opus 5 is being placed near Claude Fable 5 on capability, below Fable 5 on price, and slightly below OpenAI’s GPT-5.6 on cost.
Per million tokens, Opus 5 costs the same as Opus 4.8: $5 input / $25 output. Anthropic also said that Opus 5 is “half the price” of Fable 5.
The pricing is notable because the launch arrives amid concerns about allegedly hidden usage caps, which have led to at least one lawsuit, and broader worries about the AI money squeeze. In that context, Anthropic appears to be stressing that the new Opus model does not raise the standard Opus price.
For customers, the pitch is not only about raw model performance. It is also about whether the model can fit into daily workflows without the cost profile of Fable 5.
Fast Mode And Fallback Options
Anthropic is also giving customers a speed-focused option. Opus 5 includes “Fast mode,” a feature launching today in research preview, which offers higher speeds for double the price.
The company is also offering an option to opt in to automatic fallbacks. When safeguards require Opus 5 to decline a request, customers can choose to have the system fall back to a lower-tier model.
Those two choices reflect different customer needs. Some users may want faster responses and accept a higher price. Others may care more about continuity when a request is declined under Opus 5’s safeguards.
Together, the features show Anthropic trying to make Opus 5 useful in production settings where cost, speed, refusals, and model selection all matter. The model is not being sold only as a benchmark upgrade; it is being packaged as an enterprise tool with safety controls built into the product story.
The Bigger Signal From Anthropic
Claude Opus 5 arrives at a moment when advanced AI models are being judged on more than capability. Anthropic is tying the launch to cybersecurity safeguards, government testing, enterprise use, and pricing discipline.
The company’s message is that Opus 5 can deliver many of the benefits associated with Claude Fable 5 while remaining more suitable for everyday work. Whether customers see it that way will depend on how the model performs in complex coding tasks, knowledge work, biology, and the other domains Anthropic is highlighting.
For now, the release makes Anthropic’s model ladder clearer. Fable 5 remains the option for the most ambitious work and long-horizon AI agent projects. Opus 5 becomes the safer, lower-priced, enterprise-focused model meant to handle serious work on a daily basis.