Anthropic’s newly released Claude 3.7 Sonnet can ‘think’ as long as the user wants before giving an answer

Anthropic has introduced Claude 3.7 Sonnet, a frontier AI model designed to provide users with both immediate and more considered responses.

The company touts it as the industry’s first “hybrid AI reasoning model,” allowing users to choose how long the AI “thinks” before generating an answer.

The rollout, announced on Monday, is part of Anthropic’s broader strategy to simplify AI interactions.

Unlike traditional AI models, which require users to select from multiple versions based on cost and capability, Claude 3.7 Sonnet integrates both rapid response and deeper reasoning into a single model.

Claude 3.7 Sonnet’s premium plans only available to subscribers

Anthropic will make Claude 3.7 Sonnet available to all users and developers, though only subscribers to its premium plans will have access to the model’s reasoning capabilities.

Free-tier users will receive a non-reasoning version, which the company claims still outperforms its predecessor, Claude 3.5 Sonnet.

The model’s pricing reflects its advanced capabilities.

Claude 3.7 Sonnet is set at $3 per million input tokens and $15 per million output tokens—higher than OpenAI’s o3-mini and DeepSeek’s R1, both of which are dedicated reasoning models but lack Claude 3.7 Sonnet’s hybrid functionality.

AI reasoning gains traction in the industry

Claude 3.7 Sonnet represents a shift toward reasoning-based AI, a method many labs are adopting as traditional improvements in AI performance slow.

Competitor models, such as OpenAI’s o3-mini, Google’s Gemini 2.0 Flash Thinking, and xAI’s Grok 3 (Think), also focus on breaking problems into smaller steps to improve accuracy.

Anthropic envisions an AI that can autonomously determine how much time to spend on reasoning, eliminating the need for users to manually adjust settings.

The company’s product and research lead, Diane Penn, emphasized that reasoning should be seamlessly integrated rather than treated as a separate function.

“Just as humans use a single brain for both quick responses and deep reflection, we believe reasoning should be an integrated capability of frontier models rather than a separate model entirely. This unified approach also creates a more seamless experience for users.” Anthropic wrote in a blog post.

Competitive edge in the AI race

Claude 3.7 Sonnet’s launch comes as Anthropic seeks to strengthen its position in the competitive AI landscape, challenging OpenAI’s ChatGPT and Google’s Gemini.

The startup is reportedly in talks to raise up to $2 billion from investors including Lightspeed and Google, valuing the company at approximately $60 billion.

Amazon has already invested $8 billion in the firm.

Anthropic’s product chief, Mike Krieger, a co-founder of Instagram, sees the hybrid model as a way to streamline AI chatbot interactions.

“Models all have personalities; they’re all a bit different,” Krieger told CNBC.

“It’s a lot to ask consumers to choose the model or decide how long they want it to reason. We want people to focus on what they need, not on technical details.”

To further improve user experience, Krieger said Anthropic will introduce a feature allowing users to set a time “budget” for reasoning, ensuring efficient responses tailored to different tasks.

Additionally, the company plans to roll out a coding tool that leverages AI agents.

Industry peers follow suit

Anthropic has a track record of pioneering product innovations ahead of competitors.

The company was the first to introduce a widely available AI “agent” feature last year, a move OpenAI soon mirrored.

OpenAI CEO Sam Altman has hinted at a similar approach to simplifying AI interactions.

In a February post on X, Altman stated that OpenAI aims to eliminate the need for users to choose between models, seeking to return to a “magic unified intelligence.”

With AI labs racing to refine chatbot functionality, Claude 3.7 Sonnet’s hybrid approach positions Anthropic as a key player in shaping the future of AI reasoning and user experience.

The post Anthropic's newly released Claude 3.7 Sonnet can 'think' as long as the user wants before giving an answer appeared first on Invezz