SAN FRANCISCO — AI startup Anthropic on Friday unveiled a new artificial intelligence model called Opus 5, which the company says comes close to matching the performance of its more powerful Fable 5 model while costing users half as much.
The San Francisco-based company said the new Claude AI model is designed to handle everyday workplace tasks and computer programming work.
In an interview with Reuters, Anthropic product leader Dianne Penn described the release as a sign of how quickly the company is advancing its technology, noting that Opus 5 is more efficient than the Opus 4.8 model that came out in May.
“We’re building and continue to consistently deliver frontier intelligence and bring that as accessibly as possible with every model generation,” Penn said.
During testing, Opus 5 showed a lower ability to exploit cybersecurity weaknesses compared to Anthropic’s top-tier AI, which means the safety restrictions placed on it are less stringent than those on Fable 5. The company also said Opus 5 is harder to trick into being misused than its other current models.
Anthropic’s Fable 5, which launched in June, was briefly taken offline after U.S. officials raised concerns that its advanced capabilities could potentially be redirected toward foreign military intelligence purposes.
Penn advised users to choose Opus 5 when looking for a cost-effective option and to turn to Fable 5 for “days-long, very autonomous projects.”
When asked about the Kimi K3 “open” model from China-based Moonshot — a company the U.S. has accused of benefiting from Anthropic’s work without authorization — Penn said it “remains to be seen” how open-weight AI models perform on the kinds of complex real-world tasks that users bring to Claude.
Unlike proprietary AI systems, open-weight models give users the ability to download, run, and customize the underlying technology themselves.
