KoboldCpp
A self-contained, single-executable inference tool built on llama.cpp, running GGUF models with a built-in web UI plus image, voice, and vision features.
Strengths
- One file, zero install: a single executable with no dependencies
- A broad feature set: text, image generation, speech, vision, and an MCP server
- Backward-compatible with hundreds of GGUF model architectures
Weaknesses
- GPU acceleration needs extra SDK setup
- The Docker path is aimed at experts
- Old CPUs need workarounds
At a glance
- Licence
- AGPL-3.0
- Pricing
- Free
- Platforms
- macOS, Windows, Linux, Docker
- GPU required
- No
Links
Related guides
Entry last verified 18 August 2026.