Anthropic launches Claude 3.5 Sonnet AI model, claims it outperforms GPT-4o in some tests

TL;DR: Anthropic has been on a roll of late, playing catch-up with OpenAI’s offerings with new features. On Thursday, the company launched Claude 3.5 Sonnet, touting it as a leap over previous iterations while also claiming it beats the recently launched GPT-4 Omni on several metrics. The company also released Artifacts, a new workspace within Claude that displays things like code snippets as a separate window beside the conversations.

Claude 3.5 Sonnet is the first release of the v3.5 family, coming a few months after the release of the Claude 3 family. The biggest improvements are in the academic and coding departments. Anthropic claims the new model beats the company’s current flagship – Claude 3 Opus – in certain coding tests. “In an internal agentic coding evaluation, Claude 3.5 Sonnet solved 64% of problems, outperforming Claude 3 Opus which solved 38%,” the company noted.

The model’s also better at grasping nuance, humor, and complex instructions and has shown improvements in writing content with a natural, relatable tone. Speed improvements are on deck too, with Anthropic claiming that the new Sonnet zips along at twice the pace of Claude 3 Opus. On the visual front, Claude 3.5 Sonnet brings notable improvements in reasoning tasks like interpreting charts and graphs.

Alongside the new model, Anthropic is introducing a feature called Artifacts on Claude.ai. This addition transforms the user experience, creating a dedicated space for AI-generated content like code snippets, text documents, or website designs. Users can now view, edit, and build upon the chatbot’s creations in real time, integrating AI-generated content into their projects.

Last but not least, the model boasts a 200K token context window, giving it plenty of room to chew on complex tasks.

Claude 3.5 Sonnet is now up for grabs on Claude.ai and the iOS app, free of charge. For those with deeper pockets and higher demands, Claude Pro and Team plan subscribers can access the model with beefed-up rate limits. Anthropic has set the prices at $3 per million input tokens and $15 per million output tokens.

The improvements brought about by Claude 3.5 Sonnet will eventually make it to its other models, too, with upgraded versions of Haiku and Opus set to drop later this year.

The new release comes roughly a month after OpenAI unveiled GPT-4 Omni. While not exactly “smarter” than GPT-4, it introduced multiple industry-first features like the ability to process text, images, and audio simultaneously. It also blew people away with its unbelievably human-like conversational powers.