Instructions to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use Interchained/Muse-Spark-Open-816B-A42B-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 Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf Interchained/Muse-Spark-Open-816B-A42B-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 Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M # Run inference directly in the terminal: llama cli -hf Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
Build from source code
git clone https://github.com/ggml-org/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-app # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama serve -hf Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M # Run inference directly in the terminal: ./build/bin/llama cli -hf Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
Use Docker
docker model run hf.co/Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
- Llama new
- LM Studio
- Jan
- vLLM
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "Interchained/Muse-Spark-Open-816B-A42B-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": "Interchained/Muse-Spark-Open-816B-A42B-GGUF", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
- Ollama
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with Ollama:
ollama run hf.co/Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
- Unsloth Desktop
- Pi
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf Interchained/Muse-Spark-Open-816B-A42B-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:9931/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Docker Model Runner
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with Docker Model Runner:
docker model run hf.co/Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
- Lemonade
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
Run and chat with the model
lemonade run user.Muse-Spark-Open-816B-A42B-GGUF-Q4_K_M
List all available models
lemonade list
- Hermes Agent
How to use Interchained/Muse-Spark-Open-816B-A42B-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 Interchained/Muse-Spark-Open-816B-A42B-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://127.0.0.1:9931/v1 hermes config set model.default Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use Interchained/Muse-Spark-Open-816B-A42B-GGUF with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf Interchained/Muse-Spark-Open-816B-A42B-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://127.0.0.1:9931/v1 \ --custom-model-id "Interchained/Muse-Spark-Open-816B-A42B-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"
Muse Spark Open 816B โ Interchained GGUF
Welcome. This is Meta's Muse Spark Open 816B (816B parameters, 42B active), converted to GGUF and quantized by Interchained. Every file here is cut from official weights, and we're glad you found us.
Built with Muse Spark Open.
What you're getting
Official meta-models/Muse-Spark-Open-816B-A42B weights, pulled gated with our own access, converted with Hearth and quantized with stock llama-quantize. No third-party sources, no weight surgery. What Meta shipped is what's in these files, and we document every step so you can verify that yourself.
| File | Type | Size | Status |
|---|---|---|---|
Interchained-Muse-Spark-Open-816B-A42B-F16-00001-of-00005.gguf โฆ -00005-of-00005.gguf |
F16 master (5 shards) | ~330 GB each | โ Live |
Interchained-Muse-Spark-Open-816B-A42B-Q4_0.gguf |
Q4_0 | ~450 GB | โ Live |
Interchained-Muse-Spark-Open-816B-A42B-Q4_K_M-00001-of-00003.gguf โฆ -00003-of-00003.gguf |
Q4_K_M (3 shards) | ~249 GB ร2 + 2.4 GB | โ Live |
Interchained-Muse-Spark-Open-816B-A42B-Q5_K_M-00001-of-00002.gguf โฆ -00002-of-00002.gguf |
Q5_K_M (2 shards) | ~293 GB ร2 | โ Live |
Interchained-Muse-Spark-Open-816B-A42B-Q6_K |
Q6_K | ~740 GB | ๐จ Coming soon |
The F16 master is the source of truth behind every quant: one master, many quants, zero drift. If you want maximum fidelity, start with the F16. If you want to actually run it, the quants are the practical path.
Note: split GGUFs load transparently. Point
llama.cppat the first shard (-00001-of-00005.gguf) and it pulls in the rest automatically, same as a single file.
Provenance
- Source weights:
meta-models/Muse-Spark-Open-816B-A42B(gated, official) - Downloaded: 2026-10-16. 78 safetensors files, 1.64 TB, verified complete
- Converted by: Interchained via Hearth (
hearth convert). 1,429 tensors, F16 - Quantized by: Interchained via stock
llama-quantize(Hearth managed runtime) - License: Muse Model License. Please see
LICENSE.mdandUSE_POLICY.md
Running it
This is a datacenter-class model. The F16 master wants serious hardware; the quants are the practical way in. Load Q4_0 or Q4_K_M in llama.cpp, Ollama, LM Studio, or anything that speaks GGUF:
llama-cli -m Interchained-Muse-Spark-Open-816B-A42B-Q4_0.gguf -p "Hello, giant."
Text-only for now. No vision projector yet.
Talk to us
We'd genuinely like to hear from you. Whether you're running these files, hitting an issue, want a different quant, or just want to talk shop, come say hello:
- Website: interchained.org
- Email: dev@interchained.org or interchained@gmail.com
- Telegram: t.me/interchained
If you build something with these models, we'd love to hear about it. If something's wrong, we want to know that too. This is a front door, not a wall.
Built by Interchained. Clean conversions, no shortcuts.
- Downloads last month
- 1,289
4-bit
5-bit
16-bit
Model tree for Interchained/Muse-Spark-Open-816B-A42B-GGUF
Base model
meta-models/Muse-Spark-Open-816B-A42B