llama.cpp discloses out-of-bounds write flaw

llama.cpp's Pocket-TTS component has a heap out-of-bounds write flaw triggered by a malicious mmproj file; it requires local access and a public exploit exists. Upgrading to build b11796 fixes it, but the root-cause patch is still open.

• llama.cpp Adds Tensor Merge on GGUF Conversion2026-10-26
• 8GB GPU runs 262K-context Qwen quant at 70 T/s2026-10-24
• ISTA-DASLab releases quantized DeepSeek-V4.1-Flash2026-10-24
• Developer runs GGUF Q4 model in HTML via WebGL22026-10-23
• User tests LFM2.5-24B-A2B, finds local performance poor2026-10-22