OpenAI’s Ted Sanders explains model naming and size’s role in LLM performance
Ted Sanders, a figure at OpenAI, clarified how the company’s model naming reflects performance and size. He noted that larger models tend to be more reliable, costlier per token, and better at recalling niche facts. OpenAI’s model hierarchy—Luna, Terra, Sol, Astra—orders them by release and improvement magnitude, with version numbers like 6.1 indicating incremental updates rather than major leaps.
Key points
- Larger LLMs cost more per token but improve reliability and recall of niche facts, per OpenAI’s Ted Sanders
- OpenAI’s model hierarchy (Luna < Terra < Sol < Astra) orders releases by performance, not just size
- Version numbers like 6.1 signal incremental updates, not major breakthroughs, Sanders explained
Sanders emphasized that model performance varies across tasks, with no single metric ranking all models consistently. Version numbers also reflect shared training data or personality traits, avoiding a sequential numbering system (e.g., skipping to 6.1 instead of 5.7). The post underscores OpenAI’s approach to model iteration and transparency in its naming conventions.
Model page: Astra →
Quoting Ted Sanders
Simon Willison · 28 October 2026Loading the full article…
This text was published by Simon Willison and written by Simon Willison. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Google announces Gemini 4 Carbon, a model not yet available to users · 3 src
- Google launches Gemini 4 Carbon, claims top AI benchmark spot and cybersecurity lead · 5 src
- NVIDIA releases open-weight Nemotron 3.5 Transcribe model and Granary v2 dataset · 9 src
- OpenAI and Anthropic cut AI model prices by up to 80% amid rising competition from open-weight models · 7 src
- Ggml-org/llama.cpp merges PR #31283 adding video input for LFM3 models · 2 src
Comments
via GitHub Discussions