Large Models Now Control Computers Like Humans! Claude 3.5 Major Upgrade, Ahead of OpenAI
Claude 3.5 Sonnet's groundbreaking update lets AI control computers like humans, boosting productivity and surpassing OpenAI. Discover more here!
If you use AI daily, you'll find two common ways to use it:
Writing prompts on platforms to generate content.
Using something like Coze to manage a bunch of proxy models.
I've always thought these methods are rather cumbersome. The best way to improve efficiency with AI should be to let AI directly take over your computer.
Just a few hours ago, the Claude 3.5 model received a major update. Anthropic released an upgraded version, Claude 3.5 Sonnet, along with a new model called Claude 3.5 Haiku.
The upgraded Claude 3.5 Sonnet surpasses the previous version in all capabilities, with significant improvements in coding abilities. Claude 3.5 Haiku performs similarly to the previous largest model, Claude 3 Opus while maintaining a similar cost and speed to the previous Haiku generation.
What stands out the most is that although Claude 3.5 Sonnet hasn’t changed its version number, it has received an epic update: it can now use computers like a human!
The latest version of Claude 3.5 Sonnet can move the cursor, click on specific areas, and type through a virtual keyboard based on user instructions, mimicking human-computer interaction.
Anthropic stated that Claude 3.5 Sonnet is the first cutting-edge AI model to offer "computer usage" capabilities in a public beta test. However, they also noted that this tool is still experimental — it can be tricky and prone to errors. This functionality is currently more targeted toward developers to gather feedback.
After its release, many online users expressed their eagerness to try this feature. However, some were puzzled that such a major update didn’t come with a new model version number: "Why not change the version for such a huge update?"
In just a few hours after the release, developers have already started experimenting with Claude 3.5 Sonnet's new ability.
User Mckay Wrigley mentioned that using this feature through the API took less than 10 minutes to set up, but the possibilities it unlocks for AI are limitless, calling it a game-changer.