# Stories

> AI-drafted, editor-approved stories about what moved on the map.

- [Eleven of nineteen open language models use Apache 2.0](https://shipped.anth.us/stories/open-weights-licences/index.md): The rest split into MIT, modified MIT and four custom licences. Size does not predict which, so read the licence file. (2026-10-23)
- [Open speech models now run from 80 million to 35 billion parameters](https://shipped.anth.us/stories/open-tts-sizes-and-jobs/index.md): Open text-to-speech covers browser-sized voices, cloning and design in the middle, and a 35B model for multi-character scenes. Size tells you the job. It does not tell you the licence. (2026-10-23)
- [Decision models are starting to have family trees](https://shipped.anth.us/stories/decision-models-family-trees/index.md): Fine-tunes now stack on fine-tunes. REAX 1.1 descends from Perplexity's pplx-decider, Torchcast from StartLux, and a recommender reranker from Kev-9B. Licences and quirks travel with them. (2026-10-23)