It’s been yet another crazy few weeks in the AI world… If you blinked, you probably missed three major model releases and a side of corporate beef between, of course, OpenAI and Anthropic.
Here’s my summary of the important events:
-
The 15-Minute Showdown: Anthropic dropped Opus 4.6 yesterday and made a massive upgrade to Claude Code, only for OpenAI to retaliate with GPT 5.3 Codex exactly 15 minutes later. Anthropic allegedly bumped their launch time just to beat Sam Altman to the punch. And with rumors of an Anthropic Super Bowl ad targeting OpenAI, the “AI Wars” are clearly in full swing.
-
Anthropic’s “Agent Teams”: Anthropic is moving toward parallel productivity. Their new “Agent Teams” feature allows multiple agents to tackle different parts of big jobs simultaneously. It essentially allows complex tasks to be broken down where each agent performs their specific task instead of one agent handling it all. Not too distinct from orchestration handling on other agentic platforms, but they definitely know how to market.
-
The Million-Token Club: Opus is finally matching Google’s Gemini with a 1-million-token context window. This “massive memory” upgrade means you can dump entire codebases or a small library into a single conversation without the AI suffering from digital dementia. Though, you’re going to get the same performance trade-offs (checkout Needle-in-a-Haystack tests) when you start throwing that much context at a problem.
-
SaaS-pocalypse via PowerPoint: Claude is now living directly inside PowerPoint as a side panel. The integration is so good, it’s allegedly causing ripples in the SaaS market. I imagine it’s a scary thought for many business folk to have AI build, refine, and edit slides with little to no human input.
-
OpenAI Frontier: Not to be outdone in the enterprise space, OpenAI launched “Frontier.” It’s a platform designed to manage AI agents like human employees—complete with onboarding and feedback loops. It’s basically another agentic platform offering middle management for bots.