Good morning! Here's what's happening in AI today:
OpenAI launches GPT 5.6 with Sol, Terra, and Luna rolling out now
Notion releases Ship OS, an agent-native way to ship software
Reve releases 2.1, an upgrade to its 4K image model
How to run Fable 5 without burning through credits
And 4 new AI tools worth trying today
AI MODEL
OpenAI rolled out Sol, Terra, and Luna, its GPT 5.6 family of models, now live in ChatGPT, Codex, and the API.
Sol is the flagship model, leading in coding, knowledge work, cybersecurity, and science, positioned as OpenAI's most capable model in this release.
Terra delivers performance competitive with GPT-5.5 at a lower cost, giving developers a strong middle option for everyday tasks.
Luna is the fastest and most affordable model in the family, built for high-volume tasks where speed and cost matter most.
All three models are rolling out simultaneously across ChatGPT, Codex, and the API, meaning access is not staggered by product.
Releasing three models at once instead of one lets OpenAI cover more ground in a single announcement. Sol pushes the ceiling on capability, Terra keeps costs down without sacrificing much, and Luna makes high-volume work cheaper than before. Whichever tier you use, this is a meaningful upgrade across the board.
AI PRODUCTIVITY
Notion introduced Ship OS, an agent-native way to run your entire product development cycle inside Notion, from customer feedback to a merged pull request.
Ship OS lets you run your full product development cycle inside Notion, from the moment feedback comes in to the moment code ships.
Agents handle the triaging, routing, and summarizing work automatically, so information moves through your pipeline without manual sorting.
Your team stays focused on the judgment calls that actually need a human, while agents handle the repetitive parts of the process.
Ship OS is built to work with your existing Notion workspace rather than replacing it, so setup doesn't mean starting over.
Most product teams lose time to the busywork between feedback and shipped code, not the actual building. Ship OS automates that middle layer so your team spends more time deciding what matters and less time routing information between tools. For teams already living in Notion, this turns the workspace into an actual production pipeline.
AI TOOL
Reve released version 2.1 of its 4K image model, bringing greater prompt understanding, expanded world knowledge, and stronger foreign text rendering.
Prompt understanding has improved significantly, meaning the model interprets more complex or nuanced prompts closer to what you actually intended.
World knowledge upgrades help the model generate more accurate depictions of real objects, places, and concepts without as much manual correction.
Foreign text rendering is notably stronger in this version, addressing one of the most common failure points in AI image generation.
Reve continues to position itself around 4K output, meaning image quality stays high even as these other capabilities improve.
Text rendering and prompt accuracy have been two of the most persistent weak points in AI image tools. Reve 2.1 tackles both directly while keeping its 4K image quality intact. If you have been frustrated by garbled text or images that miss the mark on what you asked for, this update is worth testing.
HOW TO AI
You don't hire a senior architect to pour concrete. You hire them to design the building, review the plans, and sign off on the work. The crew that pours the concrete is skilled and reliable, just not who you'd want spending their time on that task.
Running everything through Fable 5 is hiring the architect to pour the concrete. It works. But you're using your most capable model for tasks a faster, more specialized model handles just as well.
The fix has a name: orchestration. Fable 5 plans. Grok 4.5, which just launched, executes. Here's the three-step setup.
Know what each model is actually for
Fable 5 is the specialist. Call it when the problem is hard enough that everyone else is at their limit. It reads a vague goal, breaks it into checkable steps, and knows which parts require judgment versus which are mechanical.
Grok 4.5 is the capable generalist. It's built for execution: writing first drafts from a clear spec, formatting, processing defined inputs, running the repetitive part of any workflow.
The rule of thumb: if the next step requires a judgment call, it stays with Fable. If it's execution against a clear spec, it goes to Grok.
Step 1: Give Fable the goal, get back a brief
Open Claude with Fable 5 selected. Describe your task, then ask it to write a detailed execution brief before producing any output, one specific enough that another model could execute it without asking questions. Save that brief.
Step 2: Hand the brief to Grok for execution
Open Grok 4.5. Paste Fable's brief exactly as written and tell it to execute the task, making reasonable assumptions on anything ambiguous and noting them rather than asking questions.
Step 3: Bring Grok's output back to Fable for review
Return to Fable with the original brief and Grok's output. Ask whether it meets the quality standard, what's missing or wrong, and whether it's ready to use. If it needs revision, get specific fix instructions and run the loop again. Most tasks close in one or two passes.
When it's still not right after two passes
Ask whether the model didn't know enough or didn't try hard enough. If Grok's execution feels shallow, tell it to redo the task with a clearer quality bar. If Fable's brief or review feels vague, tell it to go deeper. If both are genuinely trying and it's still off, the problem is upstream: the brief needs more context, not a different model.
One model thinks. One model builds. The judgment stays where it belongs, and the execution runs wherever it's fastest.
OpenAI introduced ChatGPT Work, a new agent powered by Codex and GPT 5.6 that can act across your apps and files, staying on a project for hours to turn a goal into finished work.
Higgsfield announced that Claude can now deploy Higgsfield apps in one click via Higgsfield MCP, with each app shipped with models, a database, and real-time sync built in.
Google Gemini rolled out real-time project planning in Gemini Live using Nano Banana and connected apps, letting you generate images from your camera and search nearby places just by talking, available globally at no cost.
🚀 ChatGPT: Try Sol, Terra, and Luna, OpenAI's newest model family.
🛠️ Notion: Ship software from feedback to code with Ship OS.
🖼️ Reve: Generate sharper 4K images with better prompt understanding.
🎙️ Gemini Live: Plan projects in real time using your camera, free for everyone.
THAT’S IT FOR TODAY
Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!
See you tomorrow :)
- Dr. Alvaro Cintas
✓ Full archive of premium guides with ready-to-use prompts
✓ Structured AI courses (step-by-step, start-to-finish)
✓ Every upcoming premium tutorial