# The melting temperature of a next-token distribution

> Softmax sampling is the Boltzmann distribution, so a language model can melt. I found its melting temperature in closed form and tested it on 12 models.

October 2026. Section: AI, LLMs, ML and cognitive neuroscience.

Paper: [The Melting Temperature of a Next-Token Distribution](https://doi.org/10.5281/zenodo.23219063)

## Results

- Predicted from the logit gap and log V alone, across a 976x range of vocabulary sizes
- 8 of my hypotheses died, and all 8 are published

## Story

Most of my ideas here were wrong, and the experiments said so loudly. I kept every failed hypothesis in the release because they taught me the most, and the whole thing ran on a laptop with no GPU.

## Links

- [Project](https://github.com/pragyaangaur/Token-Thermodynamics)

## See also

- [Viscous conservation laws](https://pragyaangaur.dev/work/viscous/index.md)
- [Magnetic limb-darkening audit](https://pragyaangaur.dev/work/limb-audit/index.md)
- [Cellblock](https://pragyaangaur.dev/work/cellblock/index.md)

By [Pragyaan Gaur](https://pragyaangaur.dev/). The full index is at https://pragyaangaur.dev/llms.txt and https://pragyaangaur.dev/index.json.
