Models · Updated 29 Oct, 04:53 am IST
Ted Sanders explains model size and OpenAI naming conventions
Why it matters for readers: It makes model names and numbers less confusing and clearer about what changed.
- Model performance is multidimensional but much variation is explained by model size, with larger models generally more performant and reliable.1
- Larger models tend to cost more per token and have better recall of esoteric facts.1
- OpenAI's model ordering is Luna < Terra < Sol < Astra, reflecting increasing size or capability.1
- Version numbers (for example 6.1 vs 5.6) indicate release order, similarity in training/personality, and the magnitude of improvement.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started