The llama.cpp project has released version B9406, which includes a fix for a crash related to MTP (multimodal processing) with MoE (mixture of experts) models and vision capabilities. This specific issue affected users attempting to run models like Qwen3.6-35B-A3B when processing image chunks. The update aims to resolve the GGML_ASSERT crash encountered in the get_rows function. AI
IMPACT Resolves a specific bug for users running multimodal MoE models locally, improving usability.
RANK_REASON This is a software release for an open-source project that improves functionality for running specific types of models. [lever_c_demoted from research: ic=1 ai=0.7]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →