Perplexity’s AI shopping revolution, breakthrough AI Virtual Lab, Mistral’s upgrade, and OpenAI’s platform updates — the top 4 AI news stories of the week

#1 — Perplexity transforms online shopping with AI-powered assistant and one-click checkout

Perplexity has launched a revolutionary shopping experience that combines AI-powered product discovery with seamless purchasing. The standout feature, “Buy with Pro,” enables U.S. Pro subscribers to complete purchases directly through Perplexity with a single click, including free shipping. This is complemented by “Snap to Shop,” which lets users find products by simply uploading a photo.

The platform leverages Shopify integration to access real-time product data across global merchants, delivering unbiased AI recommendations through clear, visual product cards and using straightforward language. These are not sponsored ads, but genuine recommendations. To scale this initiative, Perplexity introduced a merchant program offering retailers enhanced visibility and analytics. In a related development, Stripe released an SDK enabling AI agents to handle payments and billing, further advancing AI commerce capabilities.

#2 — The Virtual Lab: AI agents design and validate new COVID-19 treatments

A groundbreaking study introduces the Virtual Lab, where AI agents with distinct scientific specialties collaborate with human researchers to tackle complex interdisciplinary medical challenges. The system’s first major achievement: designing and validating 92 new nanobodies targeting recent SARS-CoV-2 variants.

The Virtual Lab represents a paradigm shift from AI as a tool to AI as a research partner. Through a series of structured “meetings,” AI agents specializing in chemistry, computer science, and other disciplines work together under an AI principal investigator’s guidance and with human oversight. Two of their designed nanobodies showed particularly promising results against the recent JN.1 and KP.3 variants while maintaining effectiveness against the original virus.

#3 — Mistral AI unveils major platform upgrade and releases two frontier models

Mistral AI has transformed its free web-based “Le Chat” platform with powerful new capabilities, including web search with citations, a Canvas feature for collaborative ideation, and advanced document understanding. The platform now offers integrated image generation through Black Forest Labs’ “Flux Pro,” positioning it as a comprehensive alternative to existing AI assistants.

Simultaneously, Mistral released two significant models: the 123B-parameter Mistral-Large-Instruct-2411 with a 128k context window, and Pixtral-Large-Instruct-2411, a 124B-parameter multimodal model excelling at document, chart, and image understanding. These releases, initially available as free beta features, reflect Mistral’s mission to make frontier AI capabilities widely accessible.

#4 — OpenAI brings voice to web and expands o1 model access to all API tiers

OpenAI has rolled out Advanced Voice Mode to ChatGPT’s web interface, bringing natural voice conversations to desktop users. Previously exclusive to mobile apps, this feature is now available to all paying subscribers, including Plus, Enterprise, Teams, and Edu users. The system offers nine distinct AI voices and uses GPT-4o’s audio capabilities to understand non-verbal cues (the speed of the user’s speech, for example) and respond with appropriate emotion. OpenAI plans to make this feature available to free users soon.

In a significant move for developers, OpenAI has expanded access to their o1 model family across all API tiers. Previously restricted to higher tiers, o1-preview and o1-mini are now available to all developers. The update introduces token streaming support for the first time, enabling real-time responses.

These updates reflect OpenAI’s commitment to democratizing access to their most advanced AI capabilities while continuing to innovate in natural language processing and user interaction.