A new version of the llama-runner tool automates the integration of local llama-server instances with the Claude Code agent, handling configuration, logging, and process management.
Conway Research has launched Saluki 27B, a 2-bit quantized version of Qwen3.8-27B that optimizes tool-calling performance while reducing file size to under 8 GB.
Phoronix has benchmarked the performance of Vulkan and AMD ROCm back-ends using the Lemonade local AI server on two distinct hardware platforms.
A series of experiments on a GTX 1080 Ti reveals that the 35B MoE model outperforms larger 120B and 125B counterparts in speed and coding accuracy.
A developer optimized a local server with two RTX 3060 GPUs to run a 27B parameter model, achieving a 4x speed increase for coding tasks while maintaining a 128K context window.
Nous Research has introduced a one-click local model setup for Hermes Desktop, automating hardware analysis and runtime configuration.
