💬 OpenAI Introduces Unlimited GPT 5.6 Chats

Good morning! Here's what's happening in AI today:

OpenAI unveiled GPT-5.6 Sol for Plus and Pro subscribers, delivering sharper, more factual reasoning across both Instant and deep-thinking modes, while Free and Go users get unlimited text chats with GPT-5.6 Luna.

This closes the gap between what free and paid ChatGPT users can actually do day to day. Unlimited access for Free and Go tiers removes the biggest practical friction point casual users hit, running out of messages mid-conversation. For Plus and Pro subscribers, GPT-5.6 Sol's factual accuracy gains matter most in research and coding sessions where wrong answers cost real time. It's a quiet but material shift in how much value non-paying users get from ChatGPT.

Anthropic updated Claude Fable 5's biology safeguards to cut false-positive fallbacks by about 85%, letting the model assist with a wider range of everyday health and educational questions while still routing dual-use requests, virology, toxicology, molecular design, to Opus 5.

Fewer unnecessary refusals means Fable 5 becomes noticeably more useful for students, patients, and educators asking ordinary biology or health questions. An 85% drop in false-positive fallbacks is a big swing for a safety system, suggesting the earlier version was over-blocking by a wide margin. The dual-use line stays firmly in place, so anyone doing actual research work still hits the same wall. It's a narrow but meaningful loosening, not a broad safety rollback.

Perplexity rolled out a new audio focus model for its Comet voice mode that filters out background noise, music, and other voices, making it possible to talk to the assistant clearly from a crowded coffee shop, with the feature live now for voice conversations and dictation on macOS and Windows.

Voice assistants have always struggled outside quiet rooms, and this update targets that exact failure point. Isolating a single voice from ambient noise is a harder problem than simple volume reduction, since it has to distinguish speech patterns in real time. For anyone using Perplexity on the go, this makes voice mode viable in places it wasn't before, like offices or transit. It's a small feature on paper but removes one of the more common complaints about AI voice tools generally.

You've paid for marketing judgment before, even if you called it something else. A copywriter for the homepage, an SEO consultant to tell you what's actually broken, an agency retainer just so someone stays on top of it.

That judgment can now live inside Claude Code. Install a free, open-source skill library and Claude applies the right marketing framework the moment you ask for something it covers, homepage copy, an SEO audit, a pricing strategy, an email sequence.

Get it running

Install only what you need instead:

One gotcha: asking Claude Code to run the install for you inside a chat can land it in a folder Claude Code doesn't read. Run it yourself in a terminal, or force the location:

Tell it about your business once

Skip this step and every skill re-asks the same questions forever. The product-marketing skill builds one file, .agents/product-marketing.md, holding your positioning, audience, and voice. Every other skill checks it first. Let Claude draft it from your README and existing copy, or answer its questions from scratch, and give it your customers' exact words, not polished descriptions, verbatim language converts better than anything you'd write yourself.

Just describe the task

Never call a skill by name, just describe what you need. "Write homepage copy," "set up GA4 tracking," "create a welcome sequence." Each request gets matched to the right skill automatically, and reads from the context file you already filled in.

Let OpenClaw run it without you

Everything above runs inside one Claude Code session, you ask, it works, you review, you ship. That covers most of what you need. But skills are just markdown files, any agent built on the Agent Skills spec can read them, not only Claude Code, and OpenClaw is a self-hosted agent that runs as a background service on your own machine with the same scheduled-job pattern as cron. Tell it to run the seo-audit skill every Monday and message you on Slack with what it finds, and it runs without you opening a terminal.

Windows:

Then:

That covers most of what you need, ask, it works, you review, you ship. If you want it running on a schedule without opening a terminal yourself, that's what OpenClaw is for, a self-hosted agent that runs the same skills as a background job. But none of that is required to get value out of this, Claude Code alone does the job for anything you're doing by hand. Read the full guide.

P.S. You can access all my prompts, workflows and AI trainings if you upgrade here

Cloudflare launched Kitesurf, a browser built entirely for agents that runs on Cloudflare Workers, written in Rust to use 3-7x less CPU and memory than Chromium and spin up fresh per request, now free in beta.

OpenAI introduced Agent Plugins, an open standard built with AWS, Cursor, GitHub, VS Code, and Vercel that packages Agent Skills and MCP server configurations into a shared format so a plugin built once works across compatible agent clients.

Tencent AI released Team Memory 2.0.0 in beta, a shared memory hub that turns conversations, docs, and code into four reusable assets, Chat Memory, Skill, LLM-Wiki, and Code-Graph, governed and shared across agents and frameworks, with the repo trending at #1 on GitHub's TypeScript chart.

🔍 Mintlify Index: the first publisher-direct search layer for technical knowledge, a free MCP server for agents.

🖱️ Cua Driver: extension-free browser use built into a unified computer-use driver for agents.

🧠 Ling-3.0-tiny: a 7.9B-parameter hybrid reasoning model with only 1.3B active per token, free to try now.

🎙️ Dubbing v2: sends audio or video back speaking another language in the original voice, 90+ languages, live in the API.

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