akitaonrails merged akitaonrails/llm-coding-benchmark#8 · June 18, 2026 07:44 Add Qwen 3.6 35B A3B Q8 benchmark on ik_llama.cpp #8 Adds a dedicated local profile for a dual-RTX-3090 rig running ik_llama.cpp on localhost:8000 via an OpenAI-compatible endpoint. The first RTX-3090… +377 -0 2 comments