OpenAI has dropped a couple of key ChatGPT upgrades so far this year, but neither one was the big GPT-5 upgrade we’re all waiting for. First, we got GPT-4o in May 2024 with advanced multimodal support, including Advanced Voice Mode. Then more recently, we got o1 (in preview) with more advanced reasoning capabilities.
The names make little sense, and they could be anything, really. But it’s certainly clear that OpenAI went out of its way to avoid calling either one GPT-5.
That’s understandable, considering the history of that unofficial product name. Some associated GPT-5 with AGI, the more advanced version of AI that’s capable of reasoning as well as a human, but the future of AGI is uncertain. GPT-5 also came up about a year ago, around the time of Sam Altman’s (brief) firing.
Regardless of what product names OpenAI chooses for future ChatGPT models, the next major update might be released by December. It will be different from GPT-4o and o1, and could be more powerful. But this GPT-5 candidate, reportedly called Orion, might not be available to regular users like you and me, at least not initially.
According to The Verge, OpenAI plans to launch Orion in the coming weeks, but it won’t be available through ChatGPT. Instead, Orion will be available only to the companies OpenAI works closely with. They’ll develop their own products and features built on top of Orion.
The Verge also notes that Orion is seen as the successor of GPT-4, but it’s unclear if it’ll keep the GPT-4 moniker or tick up to GPT-5.
The blog also learned that Microsoft plans to host Orion on Azure as early as November. Microsoft is one of OpenAI’s biggest partners, and its Copilot is built around ChatGPT.
Speaking of OpenAI partners, Apple integrated ChatGPT in iOS 18, though access to the chatbot is currently available only via the iOS 18.2 beta. It’s unclear whether Orion might be available to Apple initially.
When Orion does arrive, it’ll give ChatGPT a massive boost. The report notes Orion is 100 times more powerful than GPT-4, but it’s unclear what that means. It’s separate from the o1 version that OpenAI released in September, and it’s unclear whether o1’s capabilities will be integrated into Orion.
OpenAI wants to combine multiple LLMs in time to create a bigger model that might become the artificial general intelligence (AGI) product all AI companies want to develop.
Apparently, the point of o1 was, among other things, to train Orion with synthetic data. The Verge surfaced a mid-September tweet from Sam Altman that seemed to tease something big would happen in the winter. That supposedly coincided with OpenAI researchers celebrating the end of Orion’s training.
The Verge fed the cryptic post above to o1-preview, with ChatGPT concluding that Altman might be teasing Orion, the constellation that’s best visible in the night sky from November through February. The chatbot also hallucinated along the way, so there’s that.
As I said before, when looking at OpenAI ChatGPT development rumors, I’m certain that big upgrades will continue to drop. Whether GPT-4o, Advanced Voice Mode, o1/strawberry, Orion, GPT-5, or something else, OpenAI has no choice but to deliver. It can’t afford to fall behind too much, especially considering what happeend recently.
OpenAI just closed a new funding round, raising $6.6 billion in capital and agreeing to become a for-profit entity. That means it has to make (more) money. And the only way to do that is to upgrade ChatGPT with new powers.
Before this week’s report, we talked about ChatGPT Orion in early September, over a week before Altman’s tweet. At the time, The Information reported on internal OpenAI documents that brainstormed different subscription tiers for ChatGPT, including figures that went up to $2,000.
