KoboldCpp 1.116 improves local AI agent building
KoboldCpp 1.116 is useful for people running AI models on their own machines. The built-in llama.cpp now supports MCP servers, which makes it easier to connect a local model to outside tools for agent experiments.
responses are now parsed through llama.cpp's Jinja method first, which should improve out-of-the-box support for newer models that can use tools. MTP support was fixed, with better speeds expected for both Qwen and Gemma assistant MTP setups.
Download problems tied to xet speed and some SSL issues were also fixed, and reasoning content now appears correctly when streaming is turned off. The release also adds support for Krea 2 Turbo, Ideogram 4, and Boogu Edit image models, and raises the limit to 3000 characters.
Key points
- The built-in llama.cpp now supports MCP servers.
- response handling was improved for newer tool-capable models.
- MTP speed issues were fixed for Qwen and Gemma assistant setups.
- xet download speed and SSL download problems were fixed.
- Image model support expanded, and s can now be up to 3000 characters.