In partnership with

Welcome back!

Anthropic just dropped Claude Opus 4.6, pushing AI memory to new limits with a million-token context window and coordinated agent teams. The upgrade hints at a breakthrough in how models handle massive inputs with precision. Let’s unpack…

Today’s Summary:

• 🚀 Anthropic unveils Claude Opus 4.6

• 🎮 Google Genie 3 creates interactive worlds

• 💻 OpenAI launches Codex App for macOS

• 🗣️ Mistral Voxtral Transcribe 2 lowers speech costs

• 🪦 OpenAI retires GPT-4o and older ChatGPT models

• 🤖 Chrome adds Agent features for autonomous web tasks

• 🛠️ 2 new tools

TOP STORY

Anthropic releases Claude Opus 4.6 with million-token memory

The Summary: Anthropic released Claude Opus 4.6, introducing a 1M token context window for its flagship model and debuting "agent teams" that coordinate multiple AI agents on complex tasks. The model beats GPT-5.2 by 144 Elo points on real-world knowledge work and achieves 76% accuracy finding buried information in massive text volumes, versus 18.5% for its predecessor.

Key details:

• Anthropic built a full C compiler using 16 parallel Claude 4.6 agents

• Scores 1,606 Elo on GDPval-AA benchmark (finance, legal work), outperforming GPT-5.2's 1,462 and Opus 4.5's 1,416

• Agent teams split tasks across multiple AIs in parallel

• Anthropic offering $50 free credits to subscribers to try it

Why it matters: Enterprise adoption requires handling large codebases and vast document sets while pinpointing the right information. Most models collapse on large context, but Opus 4.6 performance suggests Anthropic solved something structural about attention mechanisms over big inputs. The Agent Teams feature takes this further by distributing complex tasks across specialized agents, each maintaining a focus on a subset of the problem.

System Card

FROM OUR PARTNERS

100 Prompts to Work Faster

Want to get the most out of ChatGPT?

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide

GOOGLE

Google Genie 3 creates interactive environments on demand

The Summary: Google DeepMind released Project Genie, letting users generate interactive virtual worlds from text prompts or images. The prototype runs on Genie 3, a world model that renders environments in real time as users navigate them. Users can sketch worlds, explore them for 60 seconds, and remix existing creations.

Key details:

• Generates frame-by-frame as you move, unlike static 3D scenes

• Can animate objects from real photos, such as a stuffed toy

• Sessions limited to 60 seconds because of compute costs

• First release limited to Google AI Ultra subscribers in the US

Why it matters: World models have become a new battleground for AI research. While the technology works well for gaming demos, DeepMind's real target is training embodied AI agents in simulations before deploying them to physical robots. These models remain wildly expensive to run at scale, suggesting widespread adoption is years away.

Try Project Genie

FROM OUR PARTNERS

A Smarter Way to Read the News

Tired of news that feels like noise?

Every day, 4.5 million readers turn to 1440 for their factual news fix. We sift through 100+ sources to bring you a complete summary of politics, global events, business, and culture — all in a brief 5-minute email. No spin. No slant. Just clarity.

Join for free today!

OPENAI

OpenAI launches Codex App for macOS

The Summary: OpenAI released a Codex macOS desktop app that manages multiple AI agents working simultaneously on different tasks. The app introduces automations, customizable agent personalities, and an extensible skills system for connecting to external tools. OpenAI doubled rate limits for paid users and temporarily opened access to free-tier accounts. The launch aims to compete with Claude Code's growing developer adoption.

Key details:

• Multiple agents work on the same repository without conflicts by creating isolated code copies

• Skills system demonstrated by having Codex build a game

• Free library includes skills for Figma, Linear, cloud deployment (Cloudflare, Netlify, Render, Vercel), and document creation

• Windows version coming soon

• Available now on macOS for ChatGPT Plus, Pro, Business, Enterprise, and Edu subscribers; temporarily accessible to Free and Go users

Why it matters: As developers move from single-agent coding to teams of agents across long-running tasks, the interface becomes a competitive battleground. The test isn't only on benchmark scores but whether developers trust agents enough to run completely unattended, which requires both strong model capability and effective interface design.

QUICK NEWS

Quick news

• Mistral Voxtral Transcribe 2 undercuts speech transcription costs

• OpenAI retires GPT-4o and older ChatGPT models

• Chrome gets Agent features for autonomous web tasks

TOOLS

🥇 New tools

• Supaboard - Transform data into strategy

• Apple XCode 26.3 - Introduces support for agentic coding