NE

Dory1.0 4B BF16 Llamacpp

NemotronGGUF by cloudnathan5Updated 2026-10-19
Download on Hugging Face
Parameters3.98B
Context256K
Layers72
Architecturedory
Smallest quant0.0 GB
Quick answer

Dory1.0 4B BF16 Llamacpp needs about 5.3 GB of memory at Q4_K_M (a 2.4 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 36 quants, from 0.0 GB to 8.0 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 9.7 GB.

Dory1.0 4B BF16 Llamacpp – a looped latent‑recurrent LLM

Dory1.0 4B BF16 is a 3.98 billion‑parameter looped language model from NVIDIA’s Nemotron Labs. It uses a normalized‑GPT architecture with interleaved sliding‑window and full attention, and supports multiple inference loops for compute‑accuracy trade‑offs.

Good at

  • Looped inference for efficient token usage
  • Customizable compute budgets via loop count
  • Supports last‑loop KV caching to reduce memory

Limits

  • Requires llama.cpp build with Dory support
  • No vision or multimodal capabilities
  • License governed by OpenMDW‑1.1

Best for

  • Running on GPUs with 8 GB VRAM using Q4_K_M
  • Deploying lightweight inference on CPUs with 4 GB models
  • Experimenting with loop‑based compute scaling

Who it’s for. Developers and researchers needing a flexible, looped LLM that can trade compute for accuracy. Suitable for those who can build llama.cpp with Dory support and want to experiment with different loop counts.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.0 GB 2.9 GB
Fits Get file →
IQ2_XXS1.2 GB 4.2 GB
Fits Get file →
IQ2_XS1.3 GB 4.3 GB
Fits Get file →
IQ2_S1.4 GB 4.3 GB
Fits Get file →
IQ2_M1.5 GB 4.4 GB
Fits Get file →
Q2_K_S1.5 GB 4.4 GB
Fits Get file →
Q2_K1.6 GB 4.5 GB
Fits Get file →
Q2_K1.6 GB 4.5 GB
Fits Get file →
IQ3_XXS1.7 GB 4.6 GB
Fits Get file →
IQ3_XS1.8 GB 4.7 GB
Fits Get file →
Q3_K_S1.9 GB 4.8 GB
Fits Get file →
IQ3_S1.9 GB 4.8 GB
Fits Get file →
IQ3_M1.9 GB 4.8 GB
Fits Get file →
Q2_K_L1.9 GB 4.9 GB
Fits Get file →
Q3_K_M2.0 GB 4.9 GB
Fits Get file →
Q2_K_XL2.1 GB 5.0 GB
Fits Get file →
Q3_K_L2.1 GB 5.1 GB
Fits Get file →
IQ4_XS2.2 GB 5.2 GB
Fits Get file →
—2.3 GB 5.3 GB
Fits Get file →
—2.3 GB 5.3 GB
Fits Get file →
Q4_K_S2.3 GB 5.3 GB
Fits Get file →
IQ4_NL2.3 GB 5.3 GB
Fits Get file →
Q4_K_M2.4 GB 5.3 GB
Fits Get file →
Q4_K_M2.4 GB 5.3 GB
Fits Get file →
Q3_K_XL2.6 GB 5.5 GB
Fits Get file →
Q4_K_L2.7 GB 5.6 GB
Fits Get file →
Q5_K_S2.8 GB 5.7 GB
Fits Get file →
Q5_K_M2.8 GB 5.8 GB
Fits Get file →
Q4_K_XL2.9 GB 5.8 GB
Fits Get file →
Q5_K_L3.0 GB 6.0 GB
Fits Get file →
Q6_K3.3 GB 6.2 GB
Fits Get file →
Q5_K_XL3.4 GB 6.3 GB
Fits Get file →
Q6_K_L3.4 GB 6.4 GB
Fits Get file →
Q5_K_M3.7 GB 6.6 GB
Fits Get file →
Q8_04.2 GB 7.2 GB
Fits Get file →
BF168.0 GB 10.9 GB
Too big Get file →

For 8 GB VRAM, use Q4_K_M (2.43 GB) or Q5_K_M (2.84 GB). If memory is tighter, IQ3_M (1.92 GB) or IQ2_M (1.52 GB) are viable. For maximum quality and available memory, BF16 (7.96 GB) or Q6_K (3.27 GB) can be used.

What GPU do you need for Dory1.0 4B BF16 Llamacpp?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.9 GB in use).
  • 6 GB VRAM: the best quant that fits is Q5_K_L (3.0 GB file, about 6.0 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (4.2 GB file, about 7.2 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (8.0 GB file, about 10.9 GB in use).

Which Dory1.0 4B BF16 Llamacpp quant should you choose?

Q4_K_M is the usual starting point for Dory1.0 4B BF16 Llamacpp: it balances file size and quality well. Here it is a 2.4 GB file.

The Q8_0 file is 4.2 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturedory
Parameters3.98B
Layers72
Trained context262,144 tokens
KV cache (fp16)288 KB per token

Source

Quantized bycloudnathan5
FamilyNemotron
Licenseother
Downloads (HF)2,326
Last updated2026-10-19

Frequently asked questions about Dory1.0 4B BF16 Llamacpp

How much VRAM does Dory1.0 4B BF16 Llamacpp need?

About 5.3 GB at Q4_K_M with an 8K context window. That is the 2.4 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Dory1.0 4B BF16 Llamacpp on an 8 GB GPU?

Yes, with Q8_0: the 4.2 GB file needs about 7.2 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of Dory1.0 4B BF16 Llamacpp GGUF?

It depends on the quant: file is 0.0 GB, IQ2_XXS is 1.2 GB, IQ2_XS is 1.3 GB, IQ2_S is 1.4 GB, IQ2_M is 1.5 GB, Q2_K_S is 1.5 GB and 30 more.

What context length does Dory1.0 4B BF16 Llamacpp support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.4 GB at 8K and 9.7 GB at 32K.

Where can I download Dory1.0 4B BF16 Llamacpp GGUF?

The files are on Hugging Face in the cloudnathan5/Dory1.0-4B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Dory1.0 4B BF16 Llamacpp use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Dory1.0 4B on 8 GB VRAM?

Yes, with Q4_K_M or Q5_K_M quantizations, which fit within 8 GB VRAM and use about 5.3 GB of memory at 8k context.

Does Dory1.0 support vision or multimodal input?

No, the model is purely text‑based and has no vision or multimodal capabilities.

What build of llama.cpp is required?

A llama.cpp build with Dory support (ggml-org/llama.cpp#30306) is necessary to run these GGUF files.

Other GGUFs of NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16

Similar models

Browse all →

More from cloudnathan5

All by cloudnathan5 →

More Nemotron models

All Nemotron →

Other small models

All small models →