Good morning! Here's what's happening in AI today:
OpenAI upgrades GPT Live to handle multiple tasks in one conversation
Thinking Machines releases Inkling, a new open weights model
Reve launches Reframe for expanding images in any aspect ratio
How to pick the right Claude model, not just the biggest one
And 4 new AI tools worth trying today
AI ASSISTANT
OpenAI released an upgrade to GPT Live that keeps a conversation going while handling several tasks at once.
The model can check flights, pull up local weather, and shape an itinerary in real time, all inside one ongoing conversation.
Instead of pausing between tasks, the assistant carries context forward so each new request builds on what came before.
This is aimed at real world planning moments, like sorting out travel details while still talking through the rest of a trip.
Voice assistants have historically struggled to juggle more than one task without losing the thread of the conversation. This update closes that gap, letting people plan and multitask out loud the way they would with another person. It is a meaningful step toward voice AI that feels like a real assistant instead of a command line you talk to.
AI TOOL
Thinking Machines introduced Inkling, a new model that reasons across text, image, and audio, with the full weights released openly.
Inkling works across three modalities at once, reasoning over text, images, and audio without needing separate specialized models.
The full model weights are open, letting anyone download and build on top of it rather than accessing it only through an API.
Fine tuning is available today through Tinker, giving developers a direct path to adapt the model for their own use cases.
Most multimodal models are locked behind a paid API with no way to inspect or modify how they work. Releasing Inkling with open weights gives researchers and builders a real alternative to closed models from the biggest labs. It also signals that smaller, newer labs can compete on capability while choosing a more open release strategy.
AI MODEL
Reve introduced Reframe, a new way to expand an image or zoom into its details in any aspect ratio.
Reframe can extend the edges of an existing image, filling in new space that matches the original scene naturally.
It also works in reverse, letting you crop in closer and pull more detail out of a specific part of an image.
The tool supports any aspect ratio, so an image built for one format can be reshaped for another without starting over.
Reformatting an image for a different platform or layout has always meant either cropping out important content or generating something new entirely. Reframe solves that by extending or refocusing the existing image instead of replacing it. For anyone repurposing visuals across formats, this saves real time without sacrificing what made the original image work.
HOW TO AI
I've watched more people default to the biggest, most expensive model out of habit than actually think about what the task in front of them needs. They open Claude, pick whatever sounds the most powerful, and wonder later why a two-sentence question took longer to answer than it should have.
That's not a Claude problem. That's a matching problem.
The model is a tool built for a specific kind of work, and picking the wrong one either wastes time or leaves real capability on the table. Here's how to actually choose.
Match the model to the size of the problem, not the size of the name
Haiku is built for instant answers, straightforward questions, quick lookups, simple summaries, pulling one specific fact out of a wall of text.
Sonnet is your everyday collaborator. Writing, coding, analysis, research, multi-step problems, this is where most of your daily work should sit.
Opus is the deep reasoning model, built for complex research, long technical documents, methodology critique, and agentic coding.
Fable is for your hardest work, the kind that takes planning, not just reasoning. Long-horizon tasks with many connected steps, dense source material, or a problem Opus already struggled with.
You're not locked into one setting for the whole conversation
You don't have to pick a model at the start and stay there. If Sonnet is handling your daily writing and hits something hard, switch to Opus or Fable for that one exchange, then switch back.
The same goes for effort level, the dial underneath the model. Sonnet, Opus, and Fable each have low, medium, high, and sometimes higher, which controls how much work Claude actually does, not just how long it thinks. Running a smaller model at higher effort instead of always reaching for a bigger one often holds up well.
Haiku works differently here: instead of a graduated scale, it has one toggle called Extended, which switches deep reasoning on or off.
The model you pick and the effort you set are two separate levers. Learn what each one actually does, and you stop guessing which button to press.
P.S. You can access all my prompts, workflows and AI trainings if you upgrade here
Anthropic rolled out MCP connectors for Claude Code artifacts, letting builders create dashboards and apps that fetch information and take action for each viewer on demand.
xAI open sourced Grok Build and reset usage limits for all users, giving anyone the ability to inspect and contribute to the underlying harness.
Google expanded Gemini Spark, a personal AI agent that works in the background to get things done under your direction.
🎨 Tavus Magic Canvas: Bring charts, forms, and live answers into a shared canvas.
🤖 Slack + Devin: Investigate bugs and kick off dev work from a Slack thread.
🔍 ChatGPT Search: Search across chats, projects, images, and documents in one sidebar.
📬 Notion Agent: Catch up on notifications and clear your inbox in bulk.
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