Anthropic has launched Claude Opus 5, its newest flagship AI model, positioning it as a cost-efficient alternative to its top-tier Claude Fable 5. In a press release quoted by Gizmodo, the company said Opus 5 "comes close to the frontier intelligence of Claude Fable 5 at half the price" and is available today.
The pitch is performance-per-dollar. According to Reuters, the San Francisco-based lab framed the release as an efficiency upgrade, calling the model well suited for daily office work and computer programming. Benzinga reported Anthropic says Opus 5 beats rivals on coding and business tasks, and Yahoo and Decrypt reported it outscores Fable 5 on most benchmarks despite costing less. Several outlets, including Neowin and Cryptobriefing, noted Opus 5 replaces Opus 4.8 and carries the same pricing as that earlier model. It is already available in GitHub Copilot, per The GitHub Blog.
Safety is the other headline. Mashable called it Anthropic's "safest model yet," and Axios (via Techmeme) reported the company says it "is the least susceptible to being tricked into misuse." At the same time, Engadget reported the model has less restrictive safeguards: Anthropic expects its classifiers to "intervene around 85% less often than they do for Fable 5," and Opus 5 is not covered by the company's 30-day data retention policy.
The cadence is striking. Axios noted Opus 5 is Anthropic's fourth model release in under two months, and Malay Mail framed the strategy as a bet that lower costs will win the AI race, as rivals like GPT 5.6 press on coding. Anthropic also published a public system card, which reached the front page of Hacker News.
Why it matters: as AI shifts from novelty to everyday infrastructure for coding and office tasks, the competition is moving from raw capability toward price — making powerful models cheap enough to run at scale is now as decisive as making them smart.