Quote AI & frontier models· Oct 16, 2026

New Microsoft base model beats the one powering GPT 5.6 Sol

We now have a stronger pre-trained base model than the model that currently powers GPT 5.6 Sol.

at a town hall meeting with employees

Where it comes from

Microsoft AI chief says the company beat OpenAI at its own game, using a quarter of the computing power

Business Insider · 2026-10-16

What the article says

You're currently following this author! Click to unsubscribe from email alerts. Microsoft AI CEO Mustafa Suleyman told employees an in-house model was better than a similar offering from OpenAI, one ...

Business Insider

Also from this article

Mustafa Suleyman — Microsoft trained its base model with a quarter of…Mustafa Suleyman — Microsoft's pretraining used 12 times less compute than…

More from Mustafa Suleyman

All quotes →

Microsoft trained its base model with a quarter of OpenAI's compute

“And in fact, we used one-quarter of the compute that OpenAI used to train their model.”

Chips & compute·2 weeks ago·Business Insider

Microsoft's pretraining used 12 times less compute than DeepSeek's

“12 times less computation than was used for DeepSeek, for example, on compute efficiency.”

Chips & compute·2 weeks ago·Business Insider

Nobody is matching Meta's huge AI talent packages

“I don't think anyone's matching those things.”

Talent & orgs·Dec 13·Business Insider

On the same topic: AI & frontier models

Sam Altman OpenAI

AI solving math is net good but the method matters

“As I said, I don't think it was bad for AI. And I think, on net, the world deserves to understand how the universe works. But I do have empathy for the method.”

AI & frontier models·today·Business Insider

Sam Altman OpenAI

Models now solve the hardest problems people doubted

“People were just throwing that at the hardest problems they could. And people didn't think the model was going to do these things.”

AI & frontier models·today·Business Insider

Sundar Pichai Google

Model APIs now process nearly 32 billion tokens per minute

“Our first-party model APIs are now processing nearly 32 billion tokens per minute — 40% more than last quarter and more than four times what we processed a year ago.”

AI & frontier models·yesterday·Google