Quoting Ted Sanders
Understanding how OpenAI names and versions its models helps developers and users navigate their product lineup and set expectations for model capabilities and costs.
Reported by Simon Willison ·
What happened
Ted Sanders explains the logic behind OpenAI's model naming conventions and version numbers. He notes that while language model performance is multidimensional, model size remains the primary driver of reliability, token cost, and factual recall, while version numbers reflect release order, shared traits, and the magnitude of improvement.
Who it’s about
- OpenAI
- Ted Sanders
- Simon Willison
Source
Simon Willison
Quoting Ted Sanders