Dario Amodei, leader government officer of Anthropic, on the AI Affect Summit in New Delhi, India, on Thursday, Feb. 19, 2026.
Ruhani Kaur | Bloomberg | Getty Photographs
Anthropic on Thursday introduced a brand new synthetic intelligence style, Claude Opus 4.7, which the corporate mentioned is an development over previous fashions however is “much less extensively succesful” than its most up-to-date providing, Claude Mythos Preview.
Claude Opus 4.7 is best at instrument engineering, following directions, finishing real-world paintings and is its maximum robust most often to be had style, Anthropic mentioned. However the style’s cyber functions don’t seem to be as complex as Claude Mythos Preview, which Anthropic rolled out to a choose staff of businesses as a part of a brand new cybersecurity initiative referred to as Mission Glasswing previous this month.
“We’re liberating Opus 4.7 with safeguards that routinely locate and block requests that point out prohibited or high-risk cybersecurity makes use of,” Anthropic mentioned in a free up. “What we be informed from the real-world deployment of those safeguards will assist us paintings against our eventual function of a wide free up of Mythos-class fashions.”
Since its founding in 2021, Anthropic has spent years moderately crafting its recognition as a company that is extra devoted to protected and accountable AI deployment than opponents like OpenAI.
The release of Mission Glasswing has sparked a lot of high-profile conferences between contributors of the Trump management, tech CEOs and financial institution leader executives about the protection dangers of robust AI fashions.
Anthropic does no longer plan to make Claude Mythos Preview most often to be had, however the corporate has mentioned its function is to be informed how it will sooner or later deploy Mythos-class fashions at scale.
The release of Claude Opus 4.7 on Thursday comes after Anthropic introduced Claude Opus 4.6 in February. Anthropic mentioned the brand new style outperforms Claude Opus 4.6 throughout many use instances, together with trade benchmarks for agentic coding, multidisciplinary reasoning, scaled software use and agentic laptop use, in line with a free up.
Anthropic mentioned it experimented with efforts to “differentially scale back” Claude Opus 4.7’s cyber functions all over coaching.
The corporate inspired safety pros who’re curious about the usage of the style for “official cybersecurity functions” to use via a proper verification program.
Claude Opus 4.7 is to be had throughout all of Anthropic’s Claude merchandise, its utility programming interface and thru cloud suppliers Microsoft, Google and Amazon. The brand new style is similar worth as Claude Opus 4.6, Anthropic mentioned.
