ggerganov/llama.cpp

[SYCL] Add OP im2col_3d documentation ggml merge ready SYCL

vendor : update cpp-httplib to 0.43.4 script python merge ready

feat: add MiMo v2.5 vision examples python

speculative: fix multimodal MTP seed positions model testing Nvidia GPU Vulkan examples python server ggml Apple Metal

docs: fix metrics endpoint description in server README examples server

server : fix n_predict=-2 (generate until context full) examples python server

model : fix model type check for granite/llama3 and deepseek2/glm4.7 lite model merge ready

updated nix flake nix devops merge ready

resync python

Perf audit and tweaking ggml SYCL

WebUI: make the title cool server/webui examples server

llama-quant : add missing NVFP4 default type mapping examples

opencl: add q4_1 MoE for Adreno ggml OpenCL

Mtp clean documentation model script testing android server/webui Nvidia GPU Vulkan examples python devops server ggml SYCL Apple Metal Ascend NPU WebGPU

Add preemptive priority scheduling server/webui examples python server

webui: fix LLM title generation for agentic conversations server/webui examples server

update BoringSSL to 0.20260508.0

spec : parallel drafting support examples server

hexagon: add HTP kernel for GGML_OP_GATED_DELTA_NET ggml Hexagon

convert : add split() to LoraTorchTensor in LoRA converter python merge ready

common : do not wrap raw strings in schema parser for tagged parsers

ggml: update SCHED_DEBUG output to use ggml_op_desc() ggml

CUDA: lower-case PCI bus id, standardize for ggml Nvidia GPU ggml

convert : fix RuntimeError when stripping FP8 KV-cache scales python merge ready

Feature hexagon l2 norm ggml Hexagon

Add flash attention MMA / Tiles to support MiMo-V2.5 testing Nvidia GPU python ggml

ggml-virtgpu: include missing mutex header ggml merge ready

ggml-webgpu: address precision issues for multimodal ggml WebGPU

llama : fix device state save/load ggml Apple Metal

Gemma4_26B_A4B_NvFp4 hf checkpoint convert to gguf format fixes model python