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

This issue is brought to you by Factory, the leading developer of autonomous software factories.

  • OpenAI has a big week with ChatGPT Work and new SOTA models → The company introduced a new persistent agent that merges Codex directly into the ChatGPT app. Powered by the highly praised GPT-5.6 models, including the Sol, Terra, and Luna variants, it can handle complex, hours-long workflows across apps and files while turning high-level goals into finished work. Read more here →

  • Meta releases Muse Spark 1.1, an Opus-class model at much lower cost → Meta launched a major upgrade to its multimodal reasoning model with big improvements in agentic tool use, computer control, and long-running multi-app workflows. Muse Spark 1.1 delivers performance competitive with top frontier models while being significantly cheaper and faster. Read more here →

  • SpaceXAI launches Grok 4.5 in Cursor → Cursor just added native support for Grok 4.5, giving developers full codebase context, multi-file editing, and terminal control inside the IDE. The 2-trillion-parameter model delivers Claude Opus-level performance at a much lower cost and with significantly faster responses. Read more here →

Human-like robot hands are here:

Impressive demo of the new ChatGPT Live as a language learning tool:

The official ChatGPT Live video is fun as well:

Great read on the future of AI:

  • Anthropic makes Claude Cowork available on mobile and web → Claude can now hand off tasks from a computer and continue working even when the device is off, with results accessible on mobile. Max plan users receive doubled usage limits through August, and the experience is expanding across more plans in the coming weeks. Read more here →

  • Mistral releases Robostral Navigate for embodied navigation → The company introduced an 8B model that lets robots follow natural language instructions using only a single RGB camera. It achieves state-of-the-art results on the R2R-CE benchmark while beating previous single-camera approaches by nearly 10 points, and works across wheeled, legged, and flying robots without any LiDAR or depth sensors. Read more here →

  • Tulip King highlights falling cost of local frontier models → NVIDIA’s new DGX Spark and DGX Station systems now make it possible to run 1-trillion-parameter models locally for around $94,000. The steep drop in price for serious on-prem AI hardware continues to accelerate access to private, high-performance inference. Read more here →

  • Blue Origin raises $10B at $130B valuation → Jeff Bezos’ space company is raising approximately $10 billion in its first major outside funding round, valuing it at $130 billion. The capital will support New Glenn, lunar landers, and satellite communications as Blue Origin scales up to compete more directly with SpaceX. Read more here →

  • Prime Intellect raises $130M Series A at $1B valuation → The startup closed a $130 million round led by Radical Ventures with participation from NVIDIA Ventures, Intel Capital, and Dell Technologies Capital. It is building decentralized compute, reinforcement learning frameworks, and evaluation tools so companies can train their own agentic systems instead of depending on frontier labs. Read more here →

  • Cloudflare launches Monetization Gateway with x402 payments → Cloudflare opened the waitlist for its Monetization Gateway, letting anyone charge for web pages, datasets, APIs, or MCP tools behind Cloudflare using stablecoins over the open x402 protocol. The move makes pay-per-use monetization native at the edge for both humans and AI agents. Read more here →

That’s it for this week’s Lore Brief.

See you next week!

Keep Reading