Pull Requests ggerganov/llama.cpp

jinja : implement simple sameas test testing jinja parser

#29448 opened 2026-09-25 21:13 by CISC

ggml: sycl: fix SYCL_DNNL flag report ggml SYCL

#29447 opened 2026-09-25 21:12 by a1batross

model: GraniteSpeech5ForCTC (Turbo CTC) model server mtmd conversion

#29446 opened 2026-09-25 20:57 by gabe-l-hart

jinja : support noncall test statements with arg testing jinja parser

#29443 opened 2026-09-25 20:26 by CISC

Feature: BF16/FP16 conversion to f32 chunking documentation testing devops ggml CUDA

#29442 opened 2026-09-25 19:47 by thelittlefireman

RPC: use RDMA completion channel to not spin ggml

#29440 opened 2026-09-25 18:42 by am17an

opencl: add bin kernel `kernel_gemm_noshuffle_q8_0_q8_1_dp4a_ila_a8_bin` ggml OpenCL

#29439 opened 2026-09-25 17:29 by shaofeiqi

Revert "Change max context length for auto-fitting with unified KV"

#29437 opened 2026-09-25 17:17 by gaugarg-nv

Anujj/attention scheduling stream k ggml CUDA

#29435 opened 2026-09-25 16:21 by anujj

common : show downloaded/total size in HF progress (#29318)

#29434 opened 2026-09-25 15:53 by goldnode

[Model] Support Limite 1B - Violetto model testing conversion

#29433 opened 2026-09-25 15:46 by giuseppe

tools/quantize: add --verify flag reporting per-tensor quantization error examples

#29430 opened 2026-09-25 14:40 by ceedot-rock

cuda : implement simple rule engine for selection of TOP_K implementation ggml CUDA

#29428 opened 2026-09-25 14:19 by fairydreaming

ggml: enable gfx1012 native dot instructions ggml CUDA

#29427 opened 2026-09-25 13:10 by rekk30

tests : use llama_context_ptr and --models in test-recurrent-state-rollback testing

#29426 opened 2026-09-25 12:57 by ggerganov

cuda: update uncoalesced memory reads in pool2d ggml CUDA

#29425 opened 2026-09-25 12:31 by shenron0101

ggml-cpu: enable tiled flash attention for non-vector-multiple head dims on x86 testing ggml

#29423 opened 2026-09-25 11:52 by SongXiaoXi

Fixing static analysis issues server ggml vendor

#29421 opened 2026-09-25 11:32 by apach301

llama-cli: fix broken numa flag server

#29420 opened 2026-09-25 11:04 by jimd80

[Model] Support MiniCPM-V 4.7 documentation model mtmd conversion

#29416 opened 2026-09-25 06:17 by tc-mb

RPC: Improve loading times ggml

#29413 opened 2026-09-25 05:19 by am17an

server: accept reasoning items without content in /v1/responses server

#29412 opened 2026-09-25 04:52 by shashb27

server: fixed prompt cache is lost after the server wakes up testing server

#29408 opened 2026-09-25 02:41 by willweimike

vendor : update cpp-httplib to 0.58.0 vendor

#29407 opened 2026-09-25 02:26 by cabelo

ggml-cpu: use AVX-512 VNNI+VBMI in the Q4_K 8x8 repack GEMM ggml

#29397 opened 2026-09-24 19:48 by JeremiahM37

kleidiai: use SME2 kernels for CPU flash attention testing ggml

#29394 opened 2026-09-24 18:13 by arm-bhanuarya

metal : backward RMS norm (RMS_NORM_BACK) testing ggml Apple Metal

#29389 opened 2026-09-24 16:31 by harshithkantamneni

batch: migrate speculative, mtmd and server to batch_ext server mtmd

#29385 opened 2026-09-24 15:57 by ngxson

Ggml integer overflow ggml

#29384 opened 2026-09-24 15:22 by apach301

sycl : Q5_K reorder-layout MMVQ and fused GLU ggml SYCL

#29375 opened 2026-09-24 13:34 by NickM-27