Model Half-Life
The concept of 'model half-life' suggests that the time between model releases is decreasing. An analysis of model release data from various labs indicates that while there is increased activity, the notion of halving release times every six months is misleading. Predictions about future releases are weak without sufficient data points.
- ▪The analysis covers model releases from late 2022 to the present across major US and Chinese labs.
- ▪A median gap between consecutive model releases is used to predict future drops, but this method has limitations.
- ▪The term 'model half-life' may be more of a buzzword than a precise measure of release frequency.
Opening excerpt (first ~120 words) tap to expand
model half-lifePublished by Paul Kinlan on: May 18, 2026; Reading time: 3 minutesExpand to see summaryI keep hearing people say that there is a model “half-life” which keeps dropping from years between model releases down to a few months, with the implied assumptions that model releasing will drop to even further. I’ve heard the phrase “model half-life” so much recently that I wanted to actually look at the data.I made a TSV of every headline model release from late 2022 through today across the US frontier labs (OpenAI, Anthropic, Google, xAI, Meta, Mistral) and the major Chinese labs (DeepSeek, Qwen, Zhipu, MiniMax, Moonshot, ByteDance).
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at AI Focus.
Discussion
0 commentsMore from AI Focus
