Pull Requests ggml-org/llama.cpp

Refactor llama_model_quantize_params to expose a pure C interface examples

#20346 opened 2026-03-10 14:11 by EAddario

[ggml-virtgpu] Fix some build commands of `development.md` documentation

#20341 opened 2026-03-10 09:37 by yomaytk

llama : enable chunked fused GDN path model Nvidia GPU ggml

#20340 opened 2026-03-10 08:52 by ggerganov

Fix agentic mcp image single model examples server

#20339 opened 2026-03-10 08:49 by ServeurpersoCom

vulkan: add GATED_DELTA_NET op support testing Vulkan ggml

#20334 opened 2026-03-10 08:07 by ProgenyAlpha

Model fix qwen3vl reranker support python

#20332 opened 2026-03-10 06:51 by ViniciosLugli

ggml : bump RPC version ggml

#20330 opened 2026-03-10 06:12 by ggerganov

cli : fix --reasoning-budget and --chat-template-kwargs being ignored examples

#20329 opened 2026-03-10 05:43 by TrevorS

opencl: use larger workgroup size for get_rows ggml OpenCL

#20316 opened 2026-03-09 21:03 by lhez

vulkan: partial revert #20084 Vulkan ggml

#20298 opened 2026-03-09 15:21 by jeffbolznv

Handle reasoning budget testing examples server

#20297 opened 2026-03-09 15:07 by pwilkin

vulkan: fix OOB check in flash_attn_mask_opt Vulkan ggml

#20296 opened 2026-03-09 15:07 by jeffbolznv

ci: disable coopmat on ubuntu-24-cmake-vulkan job devops

#20294 opened 2026-03-09 14:34 by 0cc4m

[SYCL] fix op ROPE, add ROPE_BACK documentation ggml SYCL

#20293 opened 2026-03-09 14:34 by arthw

Add `--force-pure-content` to force a pure content parser. examples server

#20289 opened 2026-03-09 12:32 by pwilkin

Gracefully handle undetected tool parser, print error message.

#20286 opened 2026-03-09 11:50 by pwilkin

Support refusal content for Responses API examples server

#20285 opened 2026-03-09 11:42 by pwilkin

[SYCL] fix for failed UT case: ACC, L2_NORM, UPSCALE, GEGLU ggml SYCL

#20283 opened 2026-03-09 10:21 by arthw

ggml-cuda: gdn use shared mem for HIP Nvidia GPU ggml

#20282 opened 2026-03-09 09:53 by am17an

model: add sarvam_moe architecture support model python

#20275 opened 2026-03-09 07:00 by sumitchatterjee13

CANN: handle in-place ROPE on non-contiguous f32 tensors ggml Ascend NPU

#20274 opened 2026-03-09 06:56 by noemotiovon

Read the persisted llama_kv_cell_ext for n_pos_per_embd > 1 on state_read for all sequence ids

#20273 opened 2026-03-09 06:51 by sprayandwipe

server: support chunked transfer encoding examples server

#20269 opened 2026-03-09 05:36 by crmky

WIP/POC: NVFP4 with CUDA SM120 documentation testing Nvidia GPU examples python ggml

#20247 opened 2026-03-08 18:58 by michaelw9999

metal : add Metal backend for GGML_OP_GATED_DELTA_NET ggml Apple Metal

#20244 opened 2026-03-08 18:04 by arkavo-com

gguf-py: validate metadata values against declared types python

#20242 opened 2026-03-08 16:30 by eyupcanakman

webui : add option to copy assistant response without thinking content examples server

#20238 opened 2026-03-08 15:14 by rankaiyx

Create build-apk.yml android examples

#20231 opened 2026-03-08 10:05 by subhasishlak123

ggml-webgpu: Add supports for `GGML_OP_REPEAT` documentation ggml

#20230 opened 2026-03-08 09:51 by yomaytk

Allow VisionEmbedding to recognize embedded images without loading mmproj examples server

#20228 opened 2026-03-08 08:07 by ChenYFan