Skip to content

Pull requests: ggml-org/llama.cpp

New pull request New
1,172 Open 11,961 Closed
1,172 Open 11,961 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

mtmd: use align_corners for Qwen3-VL vision position-embedding interpolation mtmd Related to multimodal functionality (video/image/audio)
#25781 opened Jul 16, 2026 by gerbenvv Loading…
3
opencl: loads quants as uint for q4_K and q5_K flat mv (optimization for Adreno A7x GPUs) ggml changes relating to the ggml tensor library for machine learning OpenCL Issues specific to the OpenCL backend
#25780 opened Jul 16, 2026 by wanghqc Contributor Loading…
CUDA: fix external compilation of q1_0 MMQ CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning
#25778 opened Jul 16, 2026 by JohannesGaessler Contributor Loading…
support MUSA flash atten for QY1+ CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning
#25776 opened Jul 16, 2026 by ways-chen Loading…
1
Add CI targets for ROCm 7.14 devops improvements to build systems and github actions
#25775 opened Jul 16, 2026 by superm1 Contributor Loading…
1
vulkan: use spec constant for matrix matrix multiplication A-type ggml changes relating to the ggml tensor library for machine learning Vulkan Issues specific to the Vulkan backend
#25773 opened Jul 16, 2026 by 0cc4m Contributor • Draft
1
opencl: do not treat NULL-mask flash attention as causal ggml changes relating to the ggml tensor library for machine learning OpenCL Issues specific to the OpenCL backend
#25771 opened Jul 16, 2026 by kumaal Loading…
1
1
server: return 400 instead of 500 on validation error with X-Conversation-Id server/ui server
#25760 opened Jul 16, 2026 by ServeurpersoCom Contributor Loading…
1
readme : add Chaty to UIs documentation Improvements or additions to documentation
#25759 opened Jul 16, 2026 by Fangyuan025 Loading…
1
metal : per-device tuned (Q, NE) for flash-attn vec Apple Metal https://en.wikipedia.org/wiki/Metal_(API) ggml changes relating to the ggml tensor library for machine learning testing Everything test related
#25750 opened Jul 16, 2026 by forforever73 Contributor Loading…
@ggerganov
sycl: single-device still requires sync ggml changes relating to the ggml tensor library for machine learning SYCL https://en.wikipedia.org/wiki/SYCL - GPU programming language
#25741 opened Jul 15, 2026 by malsbat Contributor • Draft
3
Webui bugfix for single-model mode usage server/ui
#25738 opened Jul 15, 2026 by intel00000 Loading…
2
server : add experimental MCP (stdio) server support server
#25736 opened Jul 15, 2026 by pwilkin Member Loading…
7
vulkan: clamp l_/m_ warptile WM to <= BM (fix wrong matmul on subgroupSize > 64) ggml changes relating to the ggml tensor library for machine learning Vulkan Issues specific to the Vulkan backend
#25735 opened Jul 15, 2026 by lkazarin Loading…
1
Add TML Inkling architecture conversion CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning model Model specific mtmd Related to multimodal functionality (video/image/audio) testing Everything test related
#25731 opened Jul 15, 2026 by danielhanchen Contributor • Draft
3
CUDA: Improve NVFP4 W4A4 activation quantization CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning
#25730 opened Jul 15, 2026 by ORippler Collaborator Loading…
1
llama-graph: assert that views cannot be outputs
#25729 opened Jul 15, 2026 by am17an Contributor Loading…
10
spec: add initial adaptive draft length heuristic (curently mtp only)
#25726 opened Jul 15, 2026 by Us5rName Loading…
mtmd: use RAII for setting and resetting non-causal attention mtmd Related to multimodal functionality (video/image/audio)
#25723 opened Jul 15, 2026 by nikwen Loading…
served: add support for stream: true and OpenAI fallback server
#25712 opened Jul 15, 2026 by alvarobartt Loading…
ggml-blas: disable support for hadamard matmul ggml changes relating to the ggml tensor library for machine learning
#25710 opened Jul 15, 2026 by taronaeo Member Loading…
1
Thinking sampling parametrization documentation Improvements or additions to documentation server testing Everything test related
#25709 opened Jul 15, 2026 by nmarasoiu Loading…
CUDA: add Q2_0 support CUDA Related to the CUDA backend ggml changes relating to the ggml tensor library for machine learning testing Everything test related
#25707 opened Jul 15, 2026 by khosravipasha Contributor Loading…
kleidiai : warn once when a weight type has no KleidiAI kernel ggml changes relating to the ggml tensor library for machine learning
#25701 opened Jul 15, 2026 by kkjjkamal123 Loading…
3
common : recognize 'thinking' as reasoning alias in chat diff analyzer testing Everything test related
#25695 opened Jul 15, 2026 by michaelneale Loading…
2 tasks done
7
ProTip! Exclude everything labeled bug with -label:bug.