AI startup Anthropic has unveiled its “most intelligent model” yet, Claude 3.7 Sonnet. This hybrid neural network offers both “near-instant responses” and “extended step-by-step reasoning.”
Introducing Claude 3.7 Sonnet: our most intelligent model to date. It’s a hybrid reasoning model, producing near-instant responses or extended, step-by-step thinking.
One model, two ways to think.
We’re also releasing an agentic coding tool: Claude Code. pic.twitter.com/jt7qQmFWuC
— Anthropic (@AnthropicAI) February 24, 2025
The extended reasoning mode has enhanced AI capabilities in fields such as mathematics, physics, programming, and instruction execution.
Among its features, users of the API can customize the duration of the model’s reasoning.
The developers position the neural network as a versatile tool for programming and the application of agentic tools. The team has “shifted focus to real-world tasks that better reflect user needs.”
The company has also released a tool for “agentic programming” called Claude Code, which allows AI to handle important tasks directly in the terminal.
The basic version of Claude 3.7 Sonnet will be available to all users and developers, while the extended reasoning capabilities will be exclusive to paid subscribers.
Anthropic Utilizes Pokemon for Testing
In its blog, Anthropic detailed the use of the Pokemon game for testing Claude 3.7 Sonnet. The company equipped the new model with basic memory, screen pixel input, and functions for button pressing and screen navigation, enabling it to play the game continuously.
Unlike Claude 3.0 Sonnet, which got stuck in the initial location, the new model successfully battled three Pokemon gym leaders and earned their badges.
In January, Anthropic CEO Dario Amodei announced plans to release a two-way voice mode for the Claude chatbot and a memory feature to retain more information about past conversations.
Back in December, Anthropic introduced a new artificial intelligence model, 3.5 Haiku, for users of the Claude chatbot platform.