Anthropic's latest model, Claude 3.7's performance on software engineering tasks

Hey there!

Welcome back to The Pulse, where we dive into interesting AI stories and trends backed by data, all presented through simple visuals.

> Anthropic released newest SOTA model, Claude 3.7 Sonnet

> SWE-bench verified: evaluates AI models’ ability to solve real-world software issues

> all other models score ~50%

> exceptional performance on coding tasks; majorly approved by testers across the board

> hybrid reasoning model: both instant responses + extended step-by-step thinking mode

Anthropic shares future vision for Claude. In an essay, CEO Dario Amodei claims:

> AI could compress 50-100 years of biological progress into 5-10 years

> would need new economic model as AI becomes inexpensive & extremely effective

> could enable unprecedented 20% annual GDP growth in developing regions

> with intentional effort & risk management, we could have a radically better world with AI

> gained users worldwide, not just within China

> most external traffic from Egypt, followed by the US (biggest competitor)

> other countries: Algeria (2.72%), Saudi Arabia (2.19%), Iraq (1.80%)

.