Good Morning! Here's what I have for you in today's newsletter:
OpenAI launches Ultrafast and a $500 Pro plan for 8x faster GPT-6 Astra
Google launches its new frontier model, Gemini 4 Argon
Ideogram releases Ideogram 4.5, an edit model built for repeated edits
A full motion design workflow with free tools and your existing Claude plan
4 new AI tools worth trying today
AI MODELS
OpenAI launched Ultrafast, a premium speed tier for GPT-6 Astra that reaches 300 tokens per second. It also introduced Pro 500, a $500 monthly ChatGPT plan with 25 times the Plus usage allowance.
Ultrafast runs up to 8x faster in Codex and up to 6x faster in the API. It supports GPT-6 Astra in Codex, ChatGPT Work, and the API, with GPT-6.1 Sol support coming soon.
In the API, Ultrafast costs six times the standard rate, at $60 per million input tokens and $300 per million output tokens for GPT-6 Astra.
Pro 200 subscriptions are open again with GPT-6.1 Sol access, though the plan's allowance drops to 10 times Plus. OpenAI also committed to keeping the 5-hour limit removed.
If you spend much of your day waiting for Astra in Codex, Ultrafast puts a price on getting that time back. Pro 200 also deserves a closer look: it now includes about half its previous usage allowance. Check how much you actually use before deciding whether to upgrade or renew.
AI MODELS
Google announced Gemini 4 Argon, its next-generation model, first releasing it to a small group of cybersecurity partners. Google says it is built for complex, long-running tasks in software engineering, finance, legal work and cybersecurity.
Google says the model is state of the art on certain coding and knowledge work benchmarks and beats OpenAI's GPT-6 Astra on several. It does not claim Argon wins every benchmark.
Google is raising Argon's output-token limit to 1 million tokens, up from 64K. Introductory pricing starts at $2.00 per unit and doubles to $4.00 afterward.
Trusted defenders and Google's internal teams get a version without cyber guardrails, and Google is participating in the U.S. government's voluntary pre-release process
Gemini 4 follows a long period in which Google released smaller, cheaper Flash models while OpenAI and Anthropic moved ahead at the top end. You can't use it yet: public Gemini API access and Gemini app availability are still pending, though Google says paying subscribers will get access first after more testing.
AI TOOLS
Ideogram released Ideogram 4.5, an image editing model designed to preserve pixels, colors, and textures through repeated changes. It's available in Ideogram, through its API, and on partner platforms.
You can change a product's color, swap the background, then update the copy, with the model designed to keep the rest of the image from gradually changing along the way.
High-resolution images can be edited without downsizing. The edited section is stitched back into the original at full detail, preserving sharpness for close-ups and large prints.
Generation costs up to 22 cents per image at native 2K resolution. Ideogram says it plans to release open weights, but they aren't available yet.
For ad variations or localized product shots, this could mean fewer restarts when you only need a small change. It also aims to preserve the detail needed for print. Try a few rounds of edits on your own images first; the comparisons so far come from Ideogram itself.
HOW TO AI
A complete motion design workflow using free tools and your existing Claude subscription.
Opus 5.5 writes your animation as code. A function draws each frame at a specific point in the timeline, and the rendering tools below turn those frames into a video file.
Step 1: Install the free rendering stack
brew install node ffmpeg python
pip install numpy librosa soundfile
mkdir motion-studio && cd motion-studio && npm init -y
npm i -D playwright && npx playwright install chromium
claude --model claude-opus-5-5Inside Claude Code, use /model to check the model, then /effort xhigh for the first render.
Create a CLAUDE.md file with your rendering rules (pure window.seek(t) functions, no timers or CSS transitions), style rules (layouts to avoid, one accent color), and sound rules (audio generated in code, no external API).
Step 2: Test the engine with one line
Make a dynamic 15-second motion graphics video that shows what an incredible motion designer you are, like it's your showreel for a résumé. Go all out.With those rules in place, Opus should write an index.html with a seek(t) function. Playwright captures the frames, and ffmpeg encodes the video. The first version may look generic; Step 3 gives it a style to follow.
Step 3: Build your own design system
Save five frames from motion graphics you like in an examples folder. Ask Opus to write a MOTION.md with hex colors, fonts, timing, movement, texture, and five things to avoid. Have it flag anything the still frames can’t show, such as timing, rather than guessing.
Step 4: Add the critique loop
ffmpeg -i out/final.mp4 -vf "fps=2,scale=270:-1,tile=6x5" -frames:v 1 out/contact.pngGive Opus the contact sheet and ask it to score the opening, readability, motion quality, variety, and composition from 1–10. Have it list the three biggest problems with timestamps, fix them, and render again.
Step 5: Score it, for free
If you already have a track, use python beats.py song.wav > beats.json to find the beats. Otherwise, generate clicks and whooshes directly in Node and time them to the animation. You won’t need an ElevenLabs API key.
P.S. You can access all the AI courses and trainings (including the full version of this one), prompts and workflows here.
Higgsfield launched a ChatGPT extension that brings its full interface into Codex. Powered by GPT-6.1 Sol, it includes computer use, local file access, and all of Higgsfield's automation skills to handle creative projects from start to finish.
ElevenLabs added Transcript Editing to its Speech to Text API. Add a plain-language instruction to a Scribe v2 or Scribe v2 Realtime request, and it applies the changes after transcription, returning both the edited and original transcripts.
Perplexity added interactive 3D diagrams to Computer, so the agent can build models you can explore inside the conversation, including diagrams that help you follow furniture assembly steps.
🎞️ Figma: Motion added custom animation styles and Lottie export, with audio on the timeline and text animations rolling out next week.
🎨 Tesseract: a free design agent that creates a design system with fonts, colors, and layouts, then turns it into an editable video with motion and sound.
📈 Runway: launched Runway Ads to automate the creation and production of performance ads at scale. Select enterprise partners can use it, with early access available by request.
📦 fal: added Pixelcut Looping Video, which turns one product photo into a 5 to 15 second loop at up to 1080p with a fixed camera view.
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
✓ $500+ worth of AI tools