LI

H2O Lightning 4B

LightningGGUF by h2oaiUpdated 2026-10-30
Download on Hugging Face
Parameters4.21B
Context256K
Layers33
Architectureqwen35
Smallest quant1.9 GB
Quick answer

H2O Lightning 4B needs about 4.3 GB of memory at Q4_K_M (a 2.7 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 10 quants, from 1.9 GB (IQ3_XXS) to 8.4 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.4 GB.

H2O Lightning 4B – a 4.21B‑parameter decision model for typed questions and image reasoning

H2O Lightning 4B is a 4.21‑billion‑parameter decision model built on Qwen/Qwen3.5‑4B. It answers typed decision questions and processes images, returning probabilities for choice, yes/no, score, and bounding boxes. The model is released under Apache‑2.0.

Good at

  • Typed decision answering
  • Image reasoning with bounding boxes
  • Probability‑based outputs

Limits

  • No vision support in some quantizations
  • Requires llama.cpp build 11773 or later
  • Limited to one‑way reading in default setup

Best for

  • Decision support systems
  • Document analysis with images
  • Probabilistic question answering

Who it’s for. Developers building decision‑making applications that need probability estimates for choices, true/false, and ordinal scores. The model is suitable for text‑only or image‑enabled workflows and can be run locally with llama.cpp.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ3_XXS1.9 GB 3.5 GB
Fits Get file →
IQ3_XXS2.2 GB 3.8 GB
Fits Get file →
Q4_K_M2.7 GB 4.3 GB
Fits Get file →
Q4_K_M2.7 GB 4.3 GB
Fits Get file →
Q5_K_M3.1 GB 4.7 GB
Fits Get file →
Q5_K_M3.1 GB 4.7 GB
Fits Get file →
Q8_04.5 GB 6.1 GB
Fits Get file →
Q8_04.5 GB 6.1 GB
Fits Get file →
BF168.4 GB 10.0 GB
Too big Get file →
BF168.4 GB 10.0 GB
Too big Get file →

For 8 GB VRAM, use IQ3_XXS (1.9 GB) or Q4_K_M (2.71 GB). For 16 GB VRAM, Q5_K_M (3.07 GB) or Q8_0 (4.48 GB) are suitable. BF16 (8.42 GB) requires larger memory but offers full precision.

What GPU do you need for H2O Lightning 4B?

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 IQ3_XXS (2.2 GB file, about 3.8 GB in use).
  • 6 GB VRAM: the best quant that fits is Q5_K_M (3.1 GB file, about 4.7 GB in use).
  • 8 GB VRAM: the best quant that fits is Q8_0 (4.5 GB file, about 6.1 GB in use).
  • 12 GB VRAM: the best quant that fits is BF16 (8.4 GB file, about 10.0 GB in use).

Which H2O Lightning 4B quant should you choose?

Q4_K_M is the usual starting point for H2O Lightning 4B: it balances file size and quality well. Here it is a 2.7 GB file.

The Q8_0 file is 4.5 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 IQ3_XXS at 1.9 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureqwen35
Parameters4.21B
Layers33
Trained context262,144 tokens
KV cache (fp16)132 KB per token

Source

Quantized byh2oai
FamilyLightning
Licenseapache-2.0
Downloads (HF)45,092
Last updated2026-10-30

Frequently asked questions about H2O Lightning 4B

How much VRAM does H2O Lightning 4B need?

About 4.3 GB at Q4_K_M with an 8K context window. That is the 2.7 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 H2O Lightning 4B on an 8 GB GPU?

Yes, with Q8_0: the 4.5 GB file needs about 6.1 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of H2O Lightning 4B GGUF?

It depends on the quant: IQ3_XXS is 1.9 GB, IQ3_XXS is 2.2 GB, Q4_K_M is 2.7 GB, Q4_K_M is 2.7 GB, Q5_K_M is 3.1 GB, Q5_K_M is 3.1 GB and 4 more.

What context length does H2O Lightning 4B support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.4 GB at 32K.

Where can I download H2O Lightning 4B GGUF?

The files are on Hugging Face in the h2oai/h2o-lightning-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 H2O Lightning 4B use?

Hugging Face lists the license as apache-2.0. 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 H2O Lightning 4B on 8 GB VRAM?

Yes, using IQ3_XXS or Q4_K_M quantizations, which fit within 8 GB VRAM with approximate memory usage of 3.5–4.3 GB at 8k context.

Does the model support vision tasks?

Yes, it can process images and return bounding boxes, but some quantizations may have limited vision support.

What license applies to this model?

The model is released under the Apache‑2.0 license.

Other GGUFs of h2o-lightning-4b

Similar models

Browse all →

More from h2oai

All by h2oai →

More Lightning models

All Lightning →

Other small models

All small models →