Pull Requests ggml-org/llama.cpp

Added note for compiling on integrated GPUs documentation

#18633 opened 2026-01-06 04:58 by alosslessdev

vulkan: optimize ssm_scan Vulkan ggml

#18630 opened 2026-01-05 22:39 by jeffbolznv

ggml-webgpu: Fix GGML_MEM_ALIGN to 8 for emscripten. ggml

#18628 opened 2026-01-05 16:28 by yomaytk

rpc : implement event and async backend APIs ggml

#18626 opened 2026-01-05 15:11 by rgerganov

CANN: Remove unused functions ggml Ascend NPU

#18625 opened 2026-01-05 15:10 by rauletorresc

CANN: Rename `get_env` to `get_env_as_lowercase` ggml Ascend NPU

#18624 opened 2026-01-05 14:50 by rauletorresc

sampling: add tail-free (TFS) sampling

#18612 opened 2026-01-05 06:45 by viralvgupta

Hexagon add support for f16/f32 flash attention, scale, set-rows and improve f16/32 matmul ggml

#18611 opened 2026-01-05 06:06 by max-krasnyansky

ggml webgpu: initial flashattention implementation ggml

#18610 opened 2026-01-05 05:37 by reeselevine

Fix grammar parsing issues to prevent stack overflow and hangs testing

#18604 opened 2026-01-05 02:16 by aagit

common: build as shared library when BUILD_SHARED_LIBS is ON

#18602 opened 2026-01-05 00:11 by rsauciuc

memory : add llama_memory_hybrid_iswa

#18601 opened 2026-01-04 23:23 by tdakhran

server: add group support for router mode model presets examples server

#18600 opened 2026-01-04 20:29 by ssam18

ggml: fix assertion in ggml_build_backward_expand for inplace operations ggml

#18589 opened 2026-01-04 09:37 by nlasky2000-dot

mtmd : fix integer overflow when n_tokens equals INT32_MIN examples

#18588 opened 2026-01-04 09:21 by ylwango613

ggml-backend: allow free = 0 and total = 0 to use host memory info ggml OpenCL

#18587 opened 2026-01-04 08:53 by taronaeo

Fix division by zero vulnerability in gguf_init_from_file_impl ggml

#18586 opened 2026-01-04 08:34 by ylwango613

gguf-hash: add RVV tensor hashing using xxh3 examples

#18576 opened 2026-01-04 02:11 by ixgbe

GGML RPC - Add support for Unix Domain Sockets examples ggml

#18574 opened 2026-01-03 22:44 by struct

add option --tensor-type-file to llama-quantize examples

#18572 opened 2026-01-03 21:01 by EugeoSynthesisThirtyTwo

llama: max ctx by default, fix fit magic number testing examples

#18567 opened 2026-01-03 17:37 by JohannesGaessler

cuda : check src shapes for CUDA graphs Nvidia GPU ggml

#18561 opened 2026-01-03 10:12 by ggerganov

llama-bench: Add --no-fail (-nf) option examples

#18554 opened 2026-01-02 21:03 by SamAcctX

server : add thinking content blocks to Anthropic Messages API examples python server

#18551 opened 2026-01-02 19:29 by 686f6c61

ggml : add ggml_build_forward_select model Nvidia GPU Vulkan ggml SYCL Apple Metal Ascend NPU OpenCL IBM zDNN

#18550 opened 2026-01-02 17:35 by ggerganov

graph : constant topology for tokens/embeddings inputs model

#18549 opened 2026-01-02 13:50 by ggerganov

context : reserve new scheduler when graph topology changes

#18547 opened 2026-01-02 13:31 by ggerganov

Add EXAONE MoE implementations model examples python server

#18543 opened 2026-01-02 09:36 by nuxlear

chat : add parsing for solar-open-100b testing

#18540 opened 2026-01-02 08:20 by aldehir

CUDA: cache intermediate tensors Nvidia GPU ggml

#18538 opened 2026-01-02 07:19 by am17an