Anthropic presented here new version of Claude 3.5 Sonnet.
The updated model demonstrates impressive results in testing. It outperforms GPT-4o in language understanding (GPQA, MMLU benchmarks) and math problem solving (MATH).
The highest performance of the Claude 3.5 Sonnet is shown in encoding (93.7%, HumanEval), making it the leader among all available models on the market.
Computer Usage now allows Claude to interact with interfaces as a human would: move the cursor, press buttons, and enter text.
This opens up new horizons for creating AI agents, automating routine processes and software development.
You can already try the new version of Claude 3.5 Sonnet ⤵️
Free access to
Chat GPT
and
Journey
