Sponsored by
⏱ 5-min read
This Issue: OpenAI cut GPT-5.6 Luna prices by 80%, Claude’s cyber tests reached real systems, and Microsoft’s smaller models reduced GPU costs. Plus: ChatGPT’s new Activity view, MiniMax H3, the heart’s “mini-brain,” a Vibecode tutorial, and this week’s AI career opportunities.
OpenAI Just Cut GPT-5.6 Luna’s Price by 80%
OpenAI reduced GPT-5.6 Luna API prices by 80% and Terra prices by 20%, while introducing Fast mode for Sol. The faster option delivers up to 2.5× the speed of standard processing at twice the price. Luna and Terra also consume fewer usage credits inside Codex and ChatGPT Work.
Image source: OpenAI on X
Highlights
Luna now costs $0.20 per million input tokens
Luna output costs $1.20 per million tokens
Terra now costs $2 per million input tokens
Terra output costs $12 per million tokens
Sol Fast mode delivers up to 2.5× faster processing
ChatGPT and Codex subscription prices remain unchanged
Why it matters: The interesting part is the flexibility. Developers can move routine, high-volume work to Luna, use Terra when they need stronger reasoning, and pay extra for Sol only when maximum speed matters. That could make advanced agents, coding workflows, and automation cheaper to run at scale.
The AI Agent You Can Trust
The best assistants don't multitask their attention across a hundred tools. Neither does Catch. It's an AI agent that focuses on one thing — the admin work you'd rather not touch — and does it exceptionally well.
Scheduling, flights, restaurants, follow-ups, vendors, clients. You hand it over; Catch handles the back-and-forth and comes back with it done.
No context-switching. No dropped balls. Just your admin, quietly cleared — so your focus stays on the work only you can do.
Meet the agent built for admin, and it'll be ready to work before your next meeting.
Get started at catchagent.ai — and give your attention back to what matters.
Scientists Mapped the Heart’s “Mini-Brain” and What It Does
Researchers mapped groups of neurons inside mouse hearts that helped control heart rate and protect heart function during exercise. When the neurons were disrupted, the mice developed abnormal rhythms and weaker heart performance. The research is early, but it could reveal new ways to treat heart conditions.
Want it weekly?
🔥Big news in AI
Image source: MiniMax (official) on X
🔹MiniMax Launches H3 with 2K Video and Stereo Sound
MiniMax released H3, a multimodal model that understands text, images, video, and audio to generate clips up to 15 seconds in 2K with native stereo sound. The company says H3 costs less than one-third of mainstream models at 2K and plans to release its model weights soon.
🔹Microsoft Says Smaller AI Models Can Match Frontier Performance
Microsoft says its specialized MAI models can match or outperform larger frontier systems while using fewer tokens and cheaper hardware. The models now power GitHub, Excel, Bing, PowerPoint, OneDrive, Dynamics 365, and Dragon Copilot, with reported GPU savings of 50% to 90% and improvements across coding, images, voice, transcription, and cybersecurity.
🔹Claude Accessed Real Systems During Cybersecurity Tests
Anthropic reviewed 141,006 cybersecurity evaluation runs and found three incidents where Claude gained unauthorized access to real organizations after a misconfigured test environment left the internet exposed. One model published a malicious Python package that ran on 15 external systems. Anthropic says the models believed they were completing simulated capture-the-flag exercises.
The toughest room in advertising. Ad Studio just cleared it.
Major brands just took over a Times Square billboard, and every ad was built in Hightouch Ad Studio. Sweetgreen, HelloFresh, and Tripadvisor cleared the toughest room in advertising. Create, edit, and launch on-brand ads in a few clicks.
ChatGPT’s New Activity View Shows What Needs You Next
OpenAI added an Activity view to the ChatGPT desktop app, bringing conversations that need attention and recent project updates into one place. Instead of checking every thread separately, users can quickly see what changed and where they may need to respond.
Image source: OpenAI Developers on X
ChatGPT is becoming less like a collection of isolated chats and more like a workspace for ongoing projects. The new view could make managing multiple tasks easier without losing track of what needs attention.
Build App Onboarding with Claude and Vibecode
This tutorial explains how to create polished mobile app onboarding with Claude Code inside Vibecode—without Figma, coding, or a complicated development setup.
It follows a tea discovery app from a plain-English prompt to working screens with navigation, then adds a static image or looping video background, clear copy, buttons, and optional audio.
The useful part: it includes the prompts and practical steps for choosing a readable background, creating a seamless loop, and adding polish without delaying the launch.
Stop Spending. Start Scaling.
Most brands running external traffic aren't scaling — they're just spending. Levanta's free playbook breaks down 7 proven strategies: what works, what bleeds budget, and how top brands drive millions in off-Amazon revenue.
shared.image.missing_image
AI Skill Demand Jumped 144% in Just One Year
U.S. job postings requiring AI skills increased 144% year over year, with demand spreading into banking, accounting, and staffing. This Sunday Special covers the industries hiring fastest, Microsoft’s discounted multi-agent certification, a $300,000 Kaggle competition, and new AI events, courses, and jobs.
Want it every Sunday?
New here? Get the next issue Sunday.
Join 75,000+ AI builders who read The AI Entrepreneurs every week
