Spark-X2.5 runs locally: official llama.cpp support (Windows CPU demo)

Official llama.cpp support for Spark-X2.5 is here — so you can run the models locally, no cloud required.

This short walks through a real Windows CPU demo using the community 1.7B Q4_K_M GGUF and official build b10833:

1. Download the quantized GGUF
2. Launch a local server with llama.cpp
3. Ask a question — running entirely on CPU

Both sizes are available to explore — the 4B and the 1.7B.

Get started

• 📦 Model: https://github.com/XHToken/Spark-X2.5
• ⚙️ Runtime (llama.cpp): https://github.com/ggml-org/llama.cpp

Try it locally and share your feedback, experiences, and creations — we'd love to see what you build.

SparkX25 #llamacpp #LocalAI #OpenSource #GGUF