đĻ Lemonade – v8.1.1
Hey AI Enthusiasts! đĸ
Lemonade has a fresh update â version 8.1.1 is here! This release is packed with exciting improvements for running LLMs locally with maximum performance. Check it out: https://github.com/lemonade-sdk/lemonade/releases/tag/v8.1.1
Here’s a rundown of what’s new:
- ROCm7 Support: Radeon GPU users rejoice! Now you can leverage ROCm7 as a llamacpp backend.
- NPU & Hybrid LLM Support: Updates to model builds for custom NPUs and hybrid LLMs using RAI SW 1.5.
- New GGUF Models: gpt-oss-120b-GGUF and gpt-oss-20b-GGUF are now supported in the Lemonade Server. đ
- Context Length Adjustment: Fine-tune your LLM performance with the new `–ctx-length` option in `lemonade-server serve`.
- Image Input: The LLM Chat web app now supports image input! đŧī¸
- Automated Website Publishing: Say hello to a smoother workflow with automated website publishing!
- Hot New Models: qwen3-coder and cogito-v2-109B are now available as hot GGUF models. đĨ
- Improved Model Documentation: The server\_models.md file has been updated, including information on NPU models.
- Resolved Download Issues: Large model downloads from Hugging Face are back!
Happy experimenting! đšī¸đģ
đ https://github.com/lemonade-sdk/lemonade/releases/tag/v8.1.1