# Together AI — Parcae: Doing more with fewer parameters using stable looped models

- Company: Together AI (together.ai)
- Announced: 2026-04-15T00:00:00+00:00
- Category: new-model
- Subject: Inference platform
- Models affected: Parcae
- Source: https://www.together.ai/blog/parcae
- Record: https://forck.live/items/2358-parcae-doing-more-with-fewer-parameters-using-stable-looped-models

Together AI introduces Parcae, a stable looped language model architecture that achieves the quality of a Transformer with twice the parameters, using fewer parameters by increasing recurrence. A 770M parameter Parcae matches a 1.3B parameter Transformer.

## Evidence

Verbatim from https://www.together.ai/blog/parcae:

> We present Parcae, one of the first stable architectures for looped language models, achieving the quality of a Transformer twice the size with clean, predictable training.

---

Record: https://forck.live/items/2358-parcae-doing-more-with-fewer-parameters-using-stable-looped-models
Catalogue: https://forck.live/llms.txt
Feed: https://forck.live/feed.md