The latest AI model of Anthropic spent 30 hours to code a chat app for slacks or teams. According to anthropic, it excludes about 11,000 lines of the code, and it stopped running only when the task had completed.
The model, Cloud Sonnet 4.5, was declared today, and its ability to operate autonomously for 30 hours is a large jump directly. Earlier, the company’s Opus 4 model made headlines for the ability to work for seven hours in May.
This is an important step in the battle of anthropic for the market on both AI agents and AI coding. The company said Cloud Sonnet 4.5 “the world’s best model for real -world agents, coding and use of computers” and said that it “leads the market using the computer,” computer usage features refers to anthropic debut. about a year agoAccording to anthropic, the new model specially specializes in areas such as cyber security, financial services and research. One of its beta-tasters, Canwa stated that the new model “with complex, long-term reference tasks-in-the-codbase-from engineering to in-product features and research.”
Anthropic, openi, google, and other companies are constantly releasing aggregate updates and features, which allow their technology to act as an assistant for consumers (to do research on subjects, schedule meat-ups, schedules meat-ups and see flights, and to see flights) and use (slide decks) and using flights (slide decks) Analysis of the spreadsheet). The battle for meditation and dependence gets heated almost every month, if not every week. A few days ago, Openai announced the pulse, its latest Chatgpt feature designed to be part of users’ morning routine and relevant research subjects for their days.
Anthropic also said that the new model would be combined with other updates to help developers to code their AI agents.
The company wrote in a release, “We are combining the launch of the model with access to virtual machines, memory, reference management and multi-agent support.” “It essentially package the same building block which enables developers to build their own state -of -the -art agents – power cloud code.”
Diane Penne, head of product management at Anthropic, told Ruckus In an interview that the model’s improvement in improvement capabilities in his computer also surprised him. Cloud Sonnet 4.5 is more than three times to navigate a browser and use a computer compared to anthropic technology since last October. Penn said that the team had received a response from the early customers-“The world’s Githib and Karsar”-and spent working deeply on the model last month.
Scott White, Product Lead for Cloud. Ruckus This new model operates on the “Chief-of-Staff Level” and can find availability between many people and schedules a meeting, can see a data dashboard and pull the insight together, writes the position updated based on one-one meetings with its direct reports, and much more.
Neither White nor Penn had tried to vibe-coding with new models so far Ruckus Talked to them. But Penn said that she uses Cloud Sonnet 4.5 to hire a potential new team members in anthropic.
“This is actually useful for a really continuous running signal I use, ‘search for a deep web, come up with these parameters for a source for some types of roles on your team,” said the pen. “This has really been really helpful. And I have seen that Sonnet 4.5 is better than just the past, on the quality and depth of the discoveries and really really generating a spreadsheet with LinkedI’s profile, I can email them.”
