# Ai2 — Now in Nature: Retrofitting language models to operate over bytes

- Company: Ai2 (allenai.org)
- Announced: 2026-10-07
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://allenai.org/blog/bolmo-nature
- Record: https://forck.live/items/18528-now-in-nature-retrofitting-language-models-to-operate-over-bytes
- Subject: Olmo / Molmo / Asta
- Open weights: yes
- Models affected: Bolmo 1B, Bolmo 7B, Olmo, Qwen 3 8B, Llama 3 8B, Bwen 8B, Blama 8B

Ai2's research on byteifying language models, published in Nature, demonstrates converting subword models into byte-level models with additional training. The company released new checkpoints extending the approach to Qwen 3 8B and Llama 3 8B, creating Bwen 8B and Blama 8B, and also released Stage 1 checkpoints that keep the original model frozen while training byte-level components.

## Evidence

Verbatim from https://allenai.org/blog/bolmo-nature:

> Now, the research behind Bolmo has been published in Nature, and we’re releasing new checkpoints on Hugging Face that extend the underlying approach beyond the Olmo models that served as Bolmo’s starting point to other model families.

---

Record: https://forck.live/items/18528-now-in-nature-retrofitting-language-models-to-operate-over-bytes
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md