Good morning, welcome to this week’s Lore Brief, your 3-minute brief of the most important moves in AI and tech.

  • Factory raised $150M → Factory just raised $150M at a $1.5B valuation from Keith Rabois, Shaun Maguire, Blackstone and others. Matan is one of the best founders I've met. People are going to be surprised by how big this becomes. Read more here →

  • OpenAI turns Codex into a true Mac super-app → It can now see, click, and type inside any native app, generate and iterate images without leaving your workflow, run automations that pick up where they left off, and tap into 90+ plugins for docs, project tools, and deployments. The updates start rolling out today in the Codex desktop app. Read more here →

  • Anthropic drops Claude Opus 4.7 → The upgraded model handles long-running tasks with more rigor, follows instructions precisely, and verifies its own outputs before reporting back, while tripling vision resolution to 2,576px for better agent screenshot reading. It ships today at the same price as 4.6, with new /ultrareview in Claude Code and xhigh effort mode on the API. Read more here →

  • Google DeepMind launches Gemini 3.1 Flash TTS → The most controllable text-to-speech model yet lets you direct vocal style, delivery, and pace with simple text tags, while supporting more natural speech in over 70 languages and adding SynthID watermarking on every output. It rolls out now to developers via the Gemini API and to everyone inside Google Vids. Read more here →

Interview with Anjney Midha from AMP:

Prompting Gemini 3.1 Flash TTS:

Character singing tutorial:

  • NVIDIA Research releases Lyra 2.0 → The new framework turns a single image into persistent, explorable 3D worlds that stay geometrically consistent over time and support real-time robot simulation and immersive apps, solving the “forgetting” and drifting problems of earlier models. Read more here →

  • Anthropic redesigns Claude Code desktop → You can now run multiple sessions side by side in one window with a new sidebar, plus an integrated terminal, file editing, HTML and PDF previews, and a drag-and-drop layout that keeps your CLI plugins exactly as they work on the command line. The update is live in the latest desktop app. Read more here →

  • Google ships native Gemini app for Mac → A small team built the 100% Swift app with over 100 features in under 100 days, delivering lightning-fast performance and a clean, fully native experience on Apple Silicon. Read more here →

  • Perplexity launches Personal Computer → The always-on agent runs locally on your Mac mini, works 24/7 across files, native apps like Mail and Calendar, and your browser, while you control it from your phone with secure 2FA. It rolls out today to Max subscribers and the waitlist. Read more here →

  • Alibaba open-sources Qwen3.6-35B-A3B → The sparse MoE model has 35B total parameters but only 3B active and already matches much larger models on agentic coding while delivering strong multimodal perception and reasoning. It’s fully open under Apache 2.0 and available now on Hugging Face. Read more here →

  • Anthropic engineer shares Claude Code context playbook → Thariq’s viral thread explains how to manage the new 1M context window with rewind, compact vs clear, subagents, and avoiding bad compacts so you cut token burn and keep sessions effective. Read more here →

That’s it for this week’s Lore Brief.

See you next week!

Keep Reading