# NVIDIA releases 4B‑parameter Dory model with 256K token context

> The Dory1.0‑4B model, released on Oct 16, 2026, offers looped inference and up to 256 000 token windows, available under an OpenMDW license.

Oossa · 2026-10-17 · https://oossa.com/en/nvidia-releases-4b-parameter-dory-model-with-256k-token-context

NVIDIA has added a new large language model called Dory1.0‑4B to Hugging Face. It has 4 billion parameters and can handle contexts up to 256 K tokens. The model uses a looped (latent‑recurrent) architecture that lets users trade compute for accuracy without retraining. It is offered in several formats, including BF16 and 4‑bit NVFP4, and can be used for commercial or non‑commercial projects under the OpenMDW‑1.1 license.

## The facts

- Release date: October 16, 2026 ("Release Date\nOctober 16, 2026")
- Context length: up to 256,000 tokens

## Why it matters

Developers can run a single 4B‑parameter model on varying hardware while still supporting very long prompts, reducing the need for multiple specialized models.

## Sources & references

1. [New model on Hugging Face: nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-NVFP4 (text-generation)](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-NVFP4) – NVIDIA, 2026-10-16
2. [New model on Hugging Face: nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16 (text-generation)](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16) – NVIDIA, 2026-10-16
3. [New model on Hugging Face: nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-GGUF (text-generation)](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-GGUF) – NVIDIA, 2026-10-16
4. [New model on Hugging Face: nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-MXFP8 (text-generation)](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-MXFP8) – NVIDIA, 2026-10-16

Last updated: 2026-10-17