here is the new AI that does better at all levels


Anthropic has just launched Claude Sonnet 3.5, a new version of its AI model. This doubles the speed of its predecessor and outperforms ChatGPT in several areas. Claude 3.5 offers new features that are both impressive and useful, such as an interactive window for a collaborative workspace.

Claudius 3.5 sonnet
Source: Anthropic

In the race for generative AI which started almost a year ago, companies are constantly looking to outperform their competitors. Anthropicfounded by former OpenAI executives and researchers, has just launched Claude Sonnet 3.5, a new, more powerful version of its AI model. This launch comes three months after the previous version and promises to surpass models like ChatGPT-4o and Gemini 1.5 Pro.

In this new version, Claude Sonnet 3.5 introduced a new feature called Artifacts. It is an interactive window which is adjacent to the chat window. It thus offers a collaborative work space. Users can attach files and images for Claude to interpret and use. This feature also allows you to request SVG images (vector images suitable for the web). It also allows create video games simple and other interactive applications. This innovation clearly extends the capabilities of AI beyond simple text generation.

claude sonet vs chatgptclaude sonet vs chatgpt
Source: Anthropic

Anthropic’s new AI model is more powerful and cost-effective

According to Anthropic data, Claude Sonnet 3.5 showed size improvements compared to the previous version. In internal coding tests, Claude 3.5 resolved 64% of issues, compared to 38% for Claude Opus 3.0. His creativity is closer to humans, in fact, he also obtained better results in understanding nuances, humor and complex instructions. In terms of college-level reasoning, Claude 3.5 scored 59%, compared to 53% for ChatGPT-4o. For textual reasoning, it reached 87%, outperforming it and other competing models.

Note that Claude Sonnet 3.5 is also designed to be more economical. It only costs $3 (around €2.80) per million input tokens and $15 (around €14) per million output tokens, with a context window of 200,000 tokens, or around 150,000 words or 500 pages of text. For comparison, GPT-4o costs $5 (around €4.60) per million input tokens.



Source link -101