Plus: ChatGPT, Claude, and Grok fail simultaneously; Anthropic releases AI agent blueprints for retailers
Welcome to a Friday that feels like stepping into the AGI era, even as yesterday鈥檚 outages remind us we鈥檙e still very much in it.
MODELS
馃 OpenAI launches GPT-6 Astra model
OpenAI launched GPT-6 Astra, a new frontier model, on Thursday. OpenAI President Greg Brockman said during a press briefing that the company may now be in the AGI era.
The details:
GPT-6 Astra is designed to navigate software like a human, working across browsers, spreadsheets, websites, and desktop applications.
Astra scored 72.6% on an offline subset of OSWorld 2.0, taking roughly 40 minutes per task.
OpenAI researcher Aidan Clark said Astra was the company's largest-scale training run, using more than 100,000 GPUs at Stargate.
ChatGPT, Claude, and Grok all experienced issues around the same time on Thursday. The outage affected ChatGPT logins, file uploads, voice mode, search, deep research, and image generation.
The details:
OpenAI's status page acknowledged issues across ChatGPT and Codex.
Anthropic's status page indicated Claude was experiencing elevated errors.
Anthropic's technical staff member CJ Avilla said an infrastructure issue caused a partial outage across its services.
馃洅 Anthropic releases AI agent blueprints for retailers
Anthropic released a set of blueprints for building AI shopping and merchant agents on Claude on Wednesday. Adobe Analytics reported that retail site visits driven by AI convert at a rate 60% higher than traffic from other sources.
The details:
The blueprints include a shopper-facing agent that generates preference-based suggestions and adds items to a customer's cart.
The blueprints include a merchant agent that advises on inventory decisions, pricing strategy, and marketing campaigns.
Anthropic said the merchant agent does not complete purchases on behalf of either shoppers or the retailer.
