o3-mini faster than GPT-4o-mini
In my testing o3-mini is consistently faster than gpt-4o-mini in total response time by 10-20% even as it produces more tokens.
Demo: https://imgur.com/a/o3-faster-than-gpt-4o-mini-uPQo7wK
Do we know why it's faster? And if any of those performance improvements will come to the 4o(-mini) models?
No comments yet.