Anthropic launched Claude Opus 4.7 on Thursday, its most capable generally available AI model to date, while continuing to withhold a more powerful system from the public over concerns about its potential for cybersecurity attacks.
The new model delivers a 13% lift on coding benchmarks, resolves three times more production tasks than its predecessor, and supports high-resolution image analysis up to 3.75 megapixels — roughly triple the visual capacity of Opus 4.6. Pricing holds at $5 per million input tokens and $25 per million output tokens. Opus 4.7 runs across Claude products, Anthropic’s API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.
The launch follows weeks of mounting user complaints that Opus 4.6 had quietly degraded. An AMD senior director wrote in a widely shared GitHub post that Claude had “regressed to the point it cannot be trusted to perform complex engineering.” Anthropic denied redirecting compute resources away from the model.
Sponsored · Antimony Resources
Opus 4.7 outperforms Opus 4.6, OpenAI‘s ChatGPT 5.4, and Google’s Gemini 3.1 Pro across key benchmarks — but trails Anthropic’s own Mythos Preview, a more powerful internal model the company shares only with a select group of technology and cybersecurity partners through a program called Project Glasswing.
Anthropic said the model posed too great a risk for general release and that it would test safety guardrails on less capable models first.
Read: Anthropic Withholds Powerful Claude Mythos A.I. Over Hacking Fears
Opus 4.7 is that test. Anthropic built in automatic blocks on prohibited or high-risk cybersecurity requests and launched a Cyber Verification Program for security professionals who need legitimate access.
“What we learn from the real-world deployment of these safeguards will help us work towards our eventual goal of a broad release of Mythos-class models,” the company said.
Related: Anthropic’s Mythos AI Set for Federal Deployment Amid Contract Tensions with Pentagon
Information for this story was found via the sources and companies mentioned. The author has no securities or affiliations related to the organizations discussed. Not a recommendation to buy or sell. Always do additional research and consult a professional before purchasing a security. The author holds no licenses.
