In partnership with
OpenAI didn't release GPT-5.5 Instant.
That story is fake.
So let's talk about what actually matters.
But wait, there's more:
Apple is letting you swap AI models in iOS 27
SubQ just dropped a model that can read 12 million tokens at once
I'm Alex. Welcome to L8R by Innov8.
Let's dive deep 🐰
In today's post:
OpenAI Releases GPT-5.5 Instant as New ChatGPT Default
iOS 27 Lets You Swap AI Models for Siri and System Tasks
SubQ: New LLM with 12M Context Window and Subquadratic Scaling
Fake News Alert: GPT-5.5 Instant Is Not Real
Rumors are flying around about OpenAI dropping a massive new model called GPT-5.5 Instant.
Let me stop you right there.
This model is completely fake and does not exist.
The details:
People claim this fake model cuts AI mistakes by 52.5 percent in high-stakes fields like law and finance.
OpenAI has not announced any GPT-5.5 release on their blog or social media. The real default models for ChatGPT are still GPT-4o and o1.
The source of this rumor seems to be totally made up or hallucinated by another AI.
Why it matters:
Fake AI news spreads faster than real updates.
When people invent massive leaps in tech, it confuses everyone from normal users to big investors.
You need to know what is real so you do not waste time looking for ghost tools. Stick to official OpenAI updates if you want the truth.
From our Partner
Arnold Schwarzenegger has a newsletter.
Yeah. That Arnold Schwarzenegger.
So do Codie Sanchez, Scott Galloway, Colin & Samir, Shaan Puri, and Jay Shetty. And none of them are doing it for fun. They're doing it because a list you own compounds in ways that social media never will.
beehiiv is where they built it. You can start yours for 30% off your first 3 months with code PLATFORM30. Start building today.
Apple Kicks Down Its Own Walled Garden in iOS 27
Apple is finally letting you choose your own AI.
The upcoming iOS 27 will let you swap out Apple's default models for rivals like Gemini or Claude.
This is a massive shift for a company that usually locks you into its own ecosystem.
The details:
A new framework called Extensions lets you change the AI powering Siri, Writing Tools, and Image Playground.
You can bypass Apple's native models entirely for core system tasks.
Apple is already testing this setup with Google Gemini and Anthropic Claude.
TechCrunch reported this major strategy shift on May 5, 2026.
Why it matters:
Apple loves control. They usually force you to use their tools.
Now, they are turning iOS into an open AI marketplace.
Apple's internal AI models will have to fight for your attention against the best in the business.
It means you get the best AI for your specific needs baked right into your phone.
💡 L8R's Take:
This is Apple admitting their own AI cannot win every fight.
It is a huge win for all of us.
I cannot wait to plug Claude directly into Siri and finally get smart answers..
12 Million Tokens: SubQ Breaks the Context Window
Imagine feeding an entire library into an AI in one go.
A new startup called SubQ just built an AI model that can handle 12 million tokens in a single prompt.
It uses a clever trick to keep costs low while smashing through the old limits of memory.
The details:
SubQ uses a math trick called sparse attention to process massive prompts without crashing computers.
The model boasts an insane 12-million token context window, dwarfing Google's 2-million token Gemini 1.5 Pro.
The creators claim it matches the brainpower of Claude 3 Opus 4.6.
They are also launching a coding agent and a search tool to help you dig through all that data.
Right now, the model is locked behind a private beta, so you have to request access on their website.
Why it matters:
Normal AI models get super slow and expensive when you give them long prompts. SubQ fixes this math problem.
This means you can soon drop entire codebases, massive video files, or years of legal documents into an AI without going broke.
It lowers the cost to do real, heavy work with AI.
💡 L8R's Take:
I love the big numbers, but talk is cheap until we see real tests.
If SubQ actually matches Claude 3 Opus 4.6 with 12 million tokens, it will terrify the big players.
Until I get off that waitlist, I am keeping my expectations in check..
🚀 Quick L8R Summary
GPT-5.5 Instant: This story is fake OpenAI hasn't released any model called GPT-5.5.
iOS 27 Extensions: Apple will let you swap AI models for Siri and system features instead of forcing you to use theirs.
SubQ Model: New AI handles 12 million tokens at once using sparse attention tricks..
📩 Innathe L8R engane undarunnu 👇?
We read every reply - just reply to this email and let us know how we can improve !
Appo adutha L8R il kanaam bie…👋