This was a week of model launches from all three major labs, with OpenAI's DevDay and a safety-driven reversal setting the tone.
OpenAI had the busiest week. It launched GPT-6.1 Sol, which it says nearly matches GPT-6 Astra at a lower price, and used its developer conference to show a series of announcements, including Dots, an agentic avatar. The company also abandoned plans to release an upcoming model as safety concerns escalated. That decision sits awkwardly beside the product momentum and is the story most likely to carry into next week.
Google answered with Gemini 4 Argon, which it calls its most powerful model yet. Anthropic took a different route with Claude Sonnet 5.5, positioned as a significantly cheaper and faster work partner rather than a frontier leap. Across the three, the emphasis is on cost and speed as much as raw capability.
On the enterprise side, Anthropic highlighted Barclays scaling Claude across operations and client experience. Coverage of early agentic AI adopters pointed to failed pilots and pivots, and another piece looked at how finance-focused firms are measuring ROI. Both suggest the work now is making deployments pay for themselves.
Funding and deal activity stayed brisk. ElevenLabs doubled its valuation to $22B, Armadin raised $255.5M to find vulnerabilities with swarms of AI agents, and Salesforce is reportedly acquiring customer research startup Listen Labs for about $2B.
To watch next week: whether OpenAI says more about the withheld model, and how developers respond to the cheaper tiers from OpenAI and Anthropic.