News

Research shows AI can accelerate education innovation by simulating interventions before human evaluation—potentially reducing timelines from decades to years.

GPT-4.5 achieves 62.5% accuracy on SimpleQA evaluation and reduces hallucination rates to 37.1%, compared to GPT-4o's 61.8% hallucination rate.

Salesforce's help.salesforce.com implementation handled 380,000 customer service conversations with an 84% resolution rate, with only 2% requiring humans.

Claude 3.7 Sonnet is both an ordinary LLM and a reasoning model in one: you can pick when you want the model to answer normally or think longer.

The integration enables multi-modal capabilities by incorporating Gemini's ability to process images, audio, and video alongside text with a 2M-token context window.

Meta's open-source AI models enable 26.8% higher cancer treatment prediction accuracy and reduce clinical trial matching from hundreds of days to just one day.