Anthropic has a fresh, rapid AI model and a clever fresh way to interact with chatbots

The AI ​​arms race continues: Anthropic is rolling out its latest model, called Claude 3.5 Sonnet, which it claims can equal or outperform OpenAI’s GPT-4o or Google’s Gemini on a wide range of tasks. The fresh model is now available to Claude users on the web and iOS, and Anthropic is also making it available to developers.

The Claude 3.5 Sonnet will ultimately be the middle model in the range – Anthropic uses the name Haiku for its smallest model, Sonnet for its mainstream, mid-range option and Opus for its top-of-the-line model. (The names are weird, but every AI company seems to name things in their own weird way, so let’s leave it alone.) However, the company claims that 3.5 Sonnet is better than 3 Opus, and its benchmarks show , that makes it a much rather wide margin. The fresh model is also apparently twice as rapid as the previous one, which could be an even bigger problem.

Benchmarking AI models should always be taken with a pinch of salt; there are plenty of them, it’s effortless to choose the ones that look good on you, and the styles and products change so quickly that no one seems to have much of a lead. That said, the Claude 3.5 Sonnet looks impressive: it outperformed the GPT-4o, Gemini 1.5 Pro, and Meta Llama 3 400B in seven of nine overall tests and four of five vision tests. Again, don’t read too much into this, but Anthropic seems to have created a legitimate competitor in this space.

What is this all about? Anthropic claims that Claude 3.5 Sonnet will be much better at writing and translating code, handling multi-step workflows, interpreting charts and graphs, and transcribing text from images. This fresh and improved Claude clearly has a better understanding of humor and can write in a much more human way.

Along with the fresh model, Anthropic is also introducing a fresh feature called Artifacts. With Artifacts, you can view and interact with the results of your Claude requests: if you ask a model to design something for you, they can now show you what it looks like and let you edit it directly in the app. If Claude writes you an email, you can edit it in the Claude app instead of copying it to a text editor. It’s a miniature feature, but a clever one – these AI tools need to become more than just chatbots, and features like Artifacts simply give the app more capabilities.

The artifacts actually seem to signal Claude’s long-term vision. Anthropic has long claimed to have a primary focus on enterprises (even as it employs consumer tech professionals like Instagram co-founder Mike Krieger), and in its press release announcing Claude 3.5, Sonnet stated that it plans to turn Claude into a tool for companies to “securely centralize their knowledge, documents and ongoing work in one common space.” This sounds more like Notion or Slack than ChatGPT, with Anthropic models at the center of the entire system.

For now, however, the most essential news is the model. The pace of improvement is amazing: in March, Anthropic released Claude 3 Opus, proudly claiming it was as good as GPT-4 and Gemini 1.0, before OpenAI and Google released better versions of their models. Now Anthropic has made another move, and it certainly won’t be long before its competitors do the same. Claude isn’t talked about as much as Gemini or ChatGPT, but it’s very essential in the race.