🧠 Everything Expensive Got Free This Week

ChatGPT for Academic Researchers went live on July 29. OpenAI is opening free access to 100,000 researchers through 2027, starting with roughly 10,000 seats this summer.

Approved faculty and postdocs get GPT-5.6 Sol Pro, Codex, ChatGPT Work, expanded deep research, higher usage limits and bigger context windows for 12 months. Each workspace comes with five seats, so one lead researcher can bring in four colleagues.

Access is already running at the Institute for Advanced Study and École normale supérieure. The whole thing sits inside a commitment of more than $250 million through 2027.

xAI shipped Grok Voice Think Fast 2.0 yesterday. This one powers voice agents, so think support lines, phone bots and anything where a customer talks to software instead of a person.

The interesting part is that it reasons in parallel with speech. In practice, tool calls usually fire before the agent finishes its first sentence. Time to first audio dropped from 1.25 seconds to 0.70.

Transcription got sharper too: 1.4x better than the previous version, and roughly 10x better than dedicated speech-to-text models once there's background noise, which is where real phone calls live.

It runs $0.08 per minute of audio. If you have anything pointed at grok-voice-latest, it auto-migrates on August 5.

Google is letting you create ten videos at no cost in the Gemini app, and the window closes at 11:59pm PT on August 4th. Open the tools menu and pick "Create video."

That runs on Gemini Omni, the model that replaced Veo inside the app and normally requires a paid Google AI plan. It creates, edits and remixes, so you can talk your way through changes instead of restarting from a blank prompt.

If you have been curious about paying for it, this is a free four-day audition.

Granola is the AI notetaker that sits in on your meetings and writes them up. Quill does the same job locally. One click in the menu bar records your mic and system audio as separate tracks, then transcribes both on-device into a speaker-tagged transcript. Nothing leaves your laptop, which matters if you take client calls. Needs macOS 15 or later.

Wispr Flow is the dictation app people use to talk instead of type. Parrot is the stripped-down version: hold the fn key, speak, release, and your words land at the cursor in whatever app you're in. It runs on the Apple Neural Engine, so the first word shows up in under 300ms. Needs macOS 14 or later on Apple Silicon.

Quick note: these are open-source projects on GitHub, so do your own security check before installing. They look clean from what we can tell, but your machine, your call.

Replit Design launched this week, free for everyone, and it reads your brand from your existing site instead of making you explain it.

1. Take a screenshot of your homepage, or just copy your site URL.

2. Go to replit.com/design and upload the screenshot or paste the link.

3. Describe the thing you would normally hire a designer for: an Instagram carousel, an email header, a poster, an ad creative.

4. Let it generate, then edit anything you want directly, since everything it makes stays editable.

5. Save the result as your design system so every future asset snaps to the same brand.

Here's a quick demo.

Tomorrow, Friday 31 July, the PAI team goes live to build a lead generation and outreach system in real time. You'll see exactly how to pull leads into your business and automate the follow-up.

Because we build in real-time, you get to see how we think through the problem, where we get stuck, and what the final outcome is. It's the difference between reading a recipe and watching someone cook it while talking through their decisions.

🗓️ When: Friday, July 31st @ 10AM PT (1PM ET)
 📹️ Where: Virtual