Instructions to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Ollama
How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with Ollama:
```
ollama run hf.co/ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with Docker Model Runner:
```
docker model run hf.co/ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M
```

Lemonade

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.K2-Horizon-MoVA-36B-A4B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

K2-Horizon-MoVA-36B-A4B GGUF

Community GGUF quantizations of IFM/K2-Horizon-MoVA-36B-A4B.

☕ If this GGUF made your day easier, a coffee would make mine.
Send a coffee ☕
I build and test these releases myself. Your coffee helps keep me going.
Thank you for supporting this work.

About K2-Horizon-MoVA-36B-A4B

K2-Horizon-MoVA-36B-A4B is a sparse Mixture-of-Experts model with Mixture-of-Values attention (MoVA). The upstream checkpoint has 36B total parameters and activates about 4B parameters per token. It advertises a native 524,288-token context window and is intended for reasoning, coding, agentic, and long-context workloads. See the official model card for the original model, serving recipes, prompt conventions, and upstream results.

Upstream K2-Horizon-MoVA-36B-A4B benchmark results; the image and scores are from the official model card.

Fidelity measurements

The table below compares every published GGUF file against the BF16 reference on a held-out WikiText pilot: eight chunks from wiki.test.raw and eight chunks from wiki.valid.raw, using a 4,096-token context and the same K2 llama.cpp runtime. Values are averaged across the two splits. Size (GB) uses decimal GB (file_size_bytes / 1,000,000,000). Lower Mean KLD, ΔPPL, and RMS Δp, and higher Top-1 agreement, indicate closer next-token behavior to BF16. The BF16 reference mean PPL was 6.815612 in this pilot. These are next-token fidelity measurements, not a direct percentage of capabilities retained and not a substitute for task-specific evaluation.

File	Size (GB)	Mean KLD (↓)	Top-1 vs BF16 (↑)	ΔPPL	RMS Δp
K2-Horizon-MoVA-36B-A4B-Q8_0.gguf	39.831	0.004015	97.270%	+0.071%	1.822%
K2-Horizon-MoVA-36B-A4B-Q6_K.gguf	30.765	0.005739	96.281%	+0.180%	2.191%
K2-Horizon-MoVA-36B-A4B-Q5_K_M.gguf	26.439	0.010129	95.185%	+0.527%	2.865%
K2-Horizon-MoVA-36B-A4B-Q4_K_M.gguf	22.368	0.017449	93.564%	+0.855%	3.702%
K2-Horizon-MoVA-36B-A4B-Q4_K_S.gguf	21.361	0.019009	93.454%	+0.904%	3.738%
K2-Horizon-MoVA-36B-A4B-Q4_1.gguf	23.580	0.018516	93.472%	+0.906%	3.710%
K2-Horizon-MoVA-36B-A4B-Q4_0.gguf	21.320	0.027218	92.184%	+1.244%	4.603%
K2-Horizon-MoVA-36B-A4B-IQ4_NL.gguf	21.281	0.019676	93.350%	+1.055%	3.909%
K2-Horizon-MoVA-36B-A4B-IQ4_XS.gguf	20.132	0.019835	93.558%	+0.957%	3.909%
K2-Horizon-MoVA-36B-A4B-Q3_K_L.gguf	18.841	0.047255	89.686%	+2.760%	5.925%
K2-Horizon-MoVA-36B-A4B-Q3_K_M.gguf	17.661	0.051186	89.265%	+2.920%	6.213%
K2-Horizon-MoVA-36B-A4B-IQ3_M.gguf	16.547	0.054490	89.066%	+4.063%	6.483%
K2-Horizon-MoVA-36B-A4B-IQ3_S.gguf	16.395	0.055718	89.060%	+4.034%	6.482%
K2-Horizon-MoVA-36B-A4B-Q2_K.gguf	13.622	0.116123	84.810%	+9.201%	9.529%
K2-Horizon-MoVA-36B-A4B-Q2_K_S.gguf	12.810	0.159896	82.334%	+13.465%	11.167%
K2-Horizon-MoVA-36B-A4B-IQ2_XS.gguf	11.179	0.236500	77.873%	+20.785%	13.879%
K2-Horizon-MoVA-36B-A4B-IQ1_M.gguf	8.689	0.604200	66.142%	+69.948%	22.709%
K2-Horizon-MoVA-36B-A4B-Q1_0.gguf	5.759	10.019445	1.478%	+2,069,546.308%	59.829%

Bold rows mark practical sweet spots: Q5_K_M for extra fidelity headroom, Q4_K_M as the balanced default, and IQ3_M as the compact importance-aware option. Q8_0 remains the high-bit reference profile.

In this pilot, Q1_0 shows very large divergence from the BF16 reference; the extreme-compression profiles should therefore be treated as experimental.

The machine-readable results are available in reproducibility/quality-summary.tsv, with corpus hashes, evaluation settings, and runtime provenance in reproducibility/manifest.md.

Quick start

Use a K2-Horizon-capable llama.cpp build. The explicit partial offload and CPU MoE options in this example keep the model within the available accelerator memory when full offload is not possible; adjust the GPU-layer count for your hardware.

./llama-cli \
  -m K2-Horizon-MoVA-36B-A4B-Q6_K.gguf \
  --chat-template-file reproducibility/chat_template_smoke_user.jinja \
  --jinja \
  --reasoning off \
  -p 'Answer briefly in English: What is GGUF, and why is it useful for running language models locally?' \
  -n 128 -c 4096 \
  --device CUDA0 \
  --gpu-layers 8 \
  --cpu-moe

The included template is the compatible single-turn template used by the release smoke test. The upstream full tool-aware Jinja template is not claimed as certified by this package. Long-context use may require substantially more memory than the short example above.

Reproducibility and validation

The GGUF files were quantized directly from the locked upstream BF16 input. Q8_0 was generated without an imatrix; the other ladder members used the model-specific combined importance matrix. Every published file passed the load/generate smoke test before upload. Runtime throughput data, when present, is supplementary and is recorded separately from the BF16 fidelity comparison above in reproducibility/runtime-summary.tsv.

The public package includes compact reproduction inputs and scripts. Raw conversion, imatrix, quantization, smoke-test, fidelity, and benchmark logs are kept locally under reports/ and are intentionally not uploaded.

Checksums for all published artifacts and public reproduction files are in SHA256SUMS.txt. The locked source revision, BF16 checksum, runtime commit, calibration inputs, and validation settings are in reproducibility/manifest.md.

License and attribution

The upstream model is released under the Apache License 2.0. Preserve upstream attribution and the included LICENSE when redistributing these derivative artifacts.

These are community GGUF quantizations, not an official IFM release or endorsement.

Downloads last month: 10,300

GGUF

Model size

37B params

Architecture

k2-horizon

Hardware compatibility

1-bit

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

View +1 variant

Model tree for ngquocvinh/K2-Horizon-MoVA-36B-A4B-GGUF

Base model

IFM/K2-Horizon-MoVA-36B-A4B

Quantized

(24)

this model