Pull Requests ggml-org/llama.cpp

convert : set "add bos" == True for Gemma 4 python

#21500 opened 2026-04-06 07:25 by ggerganov

fix: sanitize subprocess call in bench.py examples python server

#21498 opened 2026-04-06 06:07 by orbisai0security

vocab : remove </s> eog token if gemma4

#21492 opened 2026-04-06 01:47 by aldehir

docs: add hunyuan-ocr gguf, also add test [no ci] documentation examples

#21490 opened 2026-04-05 22:12 by ngxson

mtmd: fit_params now take into account mmproj examples server

#21489 opened 2026-04-05 21:25 by ngxson

vocab : add byte token handling to BPE detokenizer for Gemma4

#21488 opened 2026-04-05 21:08 by aldehir

console: fix stripping of \n in multiline input

#21485 opened 2026-04-05 20:51 by bipinyadav3175

llama-quantize: fix tensor-type logic

#21482 opened 2026-04-05 18:39 by theo77186

server : handle unsuccessful sink.write in chunked stream provider examples server

#21478 opened 2026-04-05 17:01 by lainon1

server: add null check for context to prevent segfault on init failure examples server

#21477 opened 2026-04-05 16:54 by Anirudh171202

gguf-py: Fix lazy tensor handling for keyword arguments python

#21476 opened 2026-04-05 16:21 by lainon1

llama-quant: use LLM_KV constants instead of hardcoded strings

#21475 opened 2026-04-05 15:28 by lainon1

CUDA: make cuda graphs props check faster Nvidia GPU ggml

#21472 opened 2026-04-05 14:13 by am17an

ggml : fix repeat_back assert with non-contiguous gradients ggml

#21467 opened 2026-04-05 11:25 by RealOrko

ggml : add GGML_OP_GATHER for DeepSeek Sparse Attention (DSA) #21149 testing ggml

#21458 opened 2026-04-05 03:55 by LilySu

vulkan: Support GGML_TYPE_NVFP4 Vulkan ggml

#21455 opened 2026-04-05 03:43 by jeffbolznv

metal : add GATED_LINEAR_ATTN op documentation testing ggml Apple Metal

#21452 opened 2026-04-05 00:58 by TheTom

Gemma 4: move some computations to BF16 model Nvidia GPU examples python ggml

#21451 opened 2026-04-05 00:24 by pwilkin

metal: speed up Qwen3-VL image encoding on large images by ~11% ggml Apple Metal

#21443 opened 2026-04-04 19:12 by Avidanborisov

eagle3: add qwen3.5 4B 9B 35B-A3B support model examples python server

#21437 opened 2026-04-04 14:35 by 36330

fix(gemma4): handle nullable type arrays testing

#21433 opened 2026-04-04 13:55 by gerstnr

vulkan: Tweak Xe2 warptile configuration Vulkan ggml

#21431 opened 2026-04-04 12:23 by TheBlueMatt

mtmd: add Gemma 4 audio conformer encoder support documentation testing Nvidia GPU examples ggml

#21421 opened 2026-04-04 09:54 by stephencox-ict

model: add Zamba2 architecture support model

#21412 opened 2026-04-04 04:58 by echo313unfolding

fix(openvino): explicit ov::Tensor frees in ggml_backend_openvino_free ggml OpenVINO

#21411 opened 2026-04-04 04:25 by thedanhoffman

cmake: add flag to use system httplib build

#21407 opened 2026-04-04 03:21 by WhyNotHugo

vendor : update cpp-httplib to 0.41.0 script python

#21405 opened 2026-04-04 02:03 by cabelo

fix tensor stride for quantized types in create_tensor

#21397 opened 2026-04-03 22:40 by Perinban

[SYCL] Add BF16 support to GET_ROWS operation ggml SYCL

#21391 opened 2026-04-03 20:26 by devedse

webui: implement pinned conversations support examples server

#21387 opened 2026-04-03 19:25 by remeh