selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

nvidia logo

NLD-Uni

nvidia·Instruct·released 5 Oct 2026

Unknown licence18.3B128K ctx ≈12 GB
Your GPU
hf download nvidia/NLD-Uni
Setup ↓

NLD-Uni is a unified multimodal model for image understanding and image generation. It combines a diffusion language model with specialized visual components and a mixture-of-transformers design.

Notify me when

How much VRAM does NLD-Uni need?

At Q4_K_M NLD-Uni needs about 12 GB with an 8K context (12 GB at Q4, 37 GB at full precision). Pick your GPU to see what fits and how fast it runs.

--vram
any
--gpus
--quant

Needs: ≈12 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.

Change quant, context or KV cache to compare
Advanced: context 8K · KV cache 16-bit
Context
KV cache

Smallest GPU for each quant and context (KV cache 16-bit)

Quantisation4K8K16K32K64K128K
Q3_K_M
Q4_K_M
Q5_K_M
Q6_K
Q8_0
FP16

≤16 GB card24–32 GB48–96 GBmore

QuantisationQualityWeightsKV cacheTotal
Q2_KHeavy lossSmallest, noticeably worse qualityHeavy loss5.6 GB 1.1 GB≈7.4 GB
Q3_K_MNoticeable lossSmall, some quality lossNoticeable loss8.3 GB 1.1 GB≈10 GB
Q4_K_MrecommendedSmall lossRecommended default for local useSmall loss10 GB 1.1 GB≈12 GB
Q5_K_MNear-losslessNear-lossless for most tasksNear-lossless12 GB 1.1 GB≈14 GB
Q6_KNear-losslessPractically losslessNear-lossless14 GB 1.1 GB≈16 GB
Q8_0LosslessLossless in practice, twice the size of Q4Lossless18 GB 1.1 GB≈21 GB
FP16 / BF16LosslessFull precision, as releasedLossless34 GB 1.1 GB≈37 GB

Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it.

How to pick a quant

Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.

  • LosslessF16, Q8_0
  • Near-losslessQ6_K, Q5_K_M
  • Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
  • Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
  • Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K

Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B

How much context fits?

Model quant
KV cache

At 8K context, Q4_K_M needs ≈12 GB(10 GB weights + 1.1 GB KV cache + 1.0 GB overhead). Pick a GPU above to see how much context fits.

01020301K2K4K8K16K32K64K128KGB

Weights + overheadKV cache

Can the Tesla T4 run NLD-Uni? →

Speed vs quality per quant

Smaller quants run faster and fit in less memory, but lose quality.

Q2_KQ3_K_MQ4_K_MQ5_K_MQ6_KQ8_0F16

Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.

Real-world speed of NLD-Uni

Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →

No medians yet. Be one of the first 3 to report a setup.

All reports

Run NLD-Uni locally

About NLD-Uni

NLD-Uni is a unified multimodal model for image understanding and image generation. It combines a diffusion language model with specialized visual components and a mixture-of-transformers design. Understanding and generation use distinct transformer parameter paths within a common multimodal framework. For image understanding, NLD-Uni encodes images into continuous visual features and generates text responses. For image generation, it models discrete visual tokens through iterative diffusion and decodes those tokens into RGB images. The architecture supports autoregressive and blockwise diffusion decoding for text, as well as image-conditioned generation and editing when enabled by the checkpoint and inference implementation. This card describes the NLD-Uni architecture generally. Install the dependencies in requirements.txt with Python 3.13 and CUDA 12.6. Install PyTorch before building FlashAttention: Validated on an NVIDIA H100 80 GB. The eight BF16 weight shards contain all 1,557 active model tensors. weights_manifest.json records their hashes. Image inputs accept Pillow images or local file paths. Image methods return Pillow images; generate returns text by default. Token IDs can be supplied through input_ids for text prompts, and return_token_ids=True returns the generated token tensor. The inference.py command exposes each task.

Full model card on Hugging Face →

Alternatives to NLD-Uni

All alternatives →
google logo

gemma-4-12B-it

google·Omni

≈11 GB

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

12B256K ctx Apache 2.0
1.6MMay 2026ollama
sensenova logo

SenseNova-U1.5-8B-MoT

sensenova·Omni

≈12 GB

SenseNova-U1.5-8B-MoT is our latest native unified multimodal checkpoint for more accurate, consistent, reliable, and aesthetically compelling visual creation.

17.5B256K ctx Apache 2.0
3.8KAug 2026gguf
google logo

gemma-4-E4B-it

google·Omni

≈5.9 GB

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

8B128K ctx Apache 2.0
4.3MMar 2026ollama
google logo

gemma-4-12B

google·Omni

≈11 GB

Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

12B256K ctx Apache 2.0
49.4KMay 2026gguf
inclusionAI logo

LLaDA2.0-Uni

inclusionAI·Omni

≈10 GB

LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model

16.3BA1.5B8K ctx Apache 2.0
2.2KApr 2026
KRAFTON logo≈22 GB

A.X K2 Raon-Speech is a bilingual English/Korean speech language model with approximately 21.2B total parameters and 3.5B active parameters.

21.2BA3B128K ctx CC BY-NC 4.0
1.5KJul 2026

NLD-Uni FAQ

How much VRAM does NLD-Uni need?
About 12 GB at Q4_K_M with an 8K context. Q8_0 needs about 21 GB and full precision about 37 GB. Longer contexts need more memory for the KV cache.
Can NLD-Uni run on the Tesla T4?
Yes. It fits at Q5_K_M (about 14 GB), at an estimated 14 tokens/s. It fits comfortably on the Arc A770 16GB, Radeon RX 7800 XT, Tesla P40.
Can I use NLD-Uni commercially?
Unclear. It is released under Unknown licence. No licence stated: assume all rights reserved. Check the licence on the model card before using it in a product.
What is the context length of NLD-Uni?
131,072 tokens (128K). The full window needs far more memory than an 8K context.

Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.