Anthropic Claims Claude 3.5 Sonnet is Best-in-Class AI Model

Anthropic is introducing a new generative AI model named Claude 3.5 Sonnet, which, while not revolutionary, marks an incremental step forward. Claude 3.5 Sonnet can analyze both text and images, and generate text, making it Anthropic's most advanced model to date, outperforming its predecessors Claude 3 Sonnet and Claude 3 Opus in various AI benchmarks for reading, coding, math, and vision.

Incremental Improvements

While benchmarks can sometimes be misleading, Claude 3.5 Sonnet has shown slight superiority over rival models like OpenAI's GPT-4o in specific tests conducted by Anthropic. In addition to improved performance, Anthropic has introduced Artifacts, a workspace where users can edit and add to content generated by their models. Currently in preview, Artifacts will soon offer features for team collaboration and knowledge base storage.

Claude 3.5 Sonnet's improvements include better understanding of nuanced instructions and concepts such as humor. More importantly for developers, it is approximately twice as fast as its predecessor, making it ideal for applications requiring prompt responses, like customer service chatbots.

Enhanced Capabilities

One of the most significant advancements in Claude 3.5 Sonnet is its enhanced ability to analyze photos. It can more accurately interpret charts and graphs and transcribe text from imperfect images, such as those with distortions. According to Michael Gerstenhaber, Anthropic’s product lead, these improvements are due to architectural tweaks and new training data, including AI-generated data.

“What matters to [businesses] is whether or not AI is helping them meet their business needs, not whether or not AI is competitive on a benchmark,” Gerstenhaber told TechCrunch.

To see these vision capabilities in action, check out the video below:

Training Data and Legal Concerns

While the specific training data remains undisclosed, it's suggested that Claude 3.5 Sonnet benefits significantly from its training sets. The secrecy may be for competitive reasons or to avoid legal challenges regarding the use of public and potentially copyrighted data.

Model Features and Availability

Claude 3.5 Sonnet has a context window of 200,000 tokens, similar to Claude 3 Sonnet, allowing it to analyze large amounts of text before generating new content. The model is available now, with free access via Anthropic's web client and the Claude iOS app. Subscribers to Anthropic's paid plans get higher rate limits, and the model is also live on Anthropic's API and platforms like Amazon Bedrock and Google Cloud's Vertex AI.

“Claude 3.5 Sonnet is really a step change in intelligence without sacrificing speed, and it sets us up for future releases along the entire Claude model family,” Gerstenhaber said.

Strategic Developments

Anthropic is focusing on building an ecosystem around its models, rather than just the models themselves. This includes experimental steering AI, integrations for model actions within apps, and tools like Artifacts. Anthropic has also expanded its presence to Europe, with offices in London and Dublin, and has brought on an Instagram co-founder as head of product.

Despite the challenges, Gerstenhaber is optimistic about the future of AI development. “There’s very rapid development and very rapid innovation, and I have no reason to believe that it’s going to slow down,” he said. You can read more about the announcement on their website.