Commits

Commit history

1. Commits on Mar 13, 2026

   • cont : minor

     ggerganov

     committed
     1794a7bView commit details

     Copy full SHA for 1794a7b

     Browse repository at this point

   • moved counter reset as per recommendation

     SoftwareRenderer

     committed
     a3050a5View commit details

     Copy full SHA for a3050a5

     Browse repository at this point

   • server: reset kill-switch on client error

     Show description for 4909dc6
     SoftwareRenderer

     committed
     4909dc6View commit details

     Copy full SHA for 4909dc6

     Browse repository at this point

   • mtmd : rename mtmd_get_audio_bitrate to mtmd_get_audio_sample_rate (#20105)

     Show description for 8f974d2
     danbev

     authored
     8f974d2View commit details

     Copy full SHA for 8f974d2

     Browse repository at this point

   • general: CONTRIBUTING.md - guidelines for quantization schemes (#19762)

     Show description for 2948e60
     

     pwilkin

     and

     JohannesGaessler

     authored
     2948e60View commit details

     Copy full SHA for 2948e60

     Browse repository at this point

   • metal : fix l2 norm scale (#20493)

     ggerganov

     authored
     73c9eb8View commit details

     Copy full SHA for 73c9eb8

     Browse repository at this point

   • convert : fix/suppress pyright errors (#20442)

     Show description for 983df14
     danbev

     authored
     983df14View commit details

     Copy full SHA for 983df14

     Browse repository at this point
2. Commits on Mar 12, 2026

   • llama : disable graph reuse with pipeline parallelism (#20463)

     ggerganov

     authored
     57819b8View commit details

     Copy full SHA for 57819b8

     Browse repository at this point

   • vendor : update cpp-httplib to 0.37.1 (#20390)

     cabelo

     authored
     557fe2dView commit details

     Copy full SHA for 557fe2d

     Browse repository at this point

   • tests : use `reasoning` instead of `reasoning_budget` in server tests (#20432)

     pwilkin

     authored
     0e81041View commit details

     Copy full SHA for 0e81041

     Browse repository at this point

   • test-backend-ops: allow loading tests from file and parsing model operators into file (#19896)

     Show description for 128142f
     0cc4m

     authored
     128142fView commit details

     Copy full SHA for 128142f

     Browse repository at this point

   • common : update completion executables list [no ci] (#19934)

     Show description for 6de1bc6
     danbev

     authored
     6de1bc6View commit details

     Copy full SHA for 6de1bc6

     Browse repository at this point

   • grammar: Fix grammar root symbol check (#19761)

     Show description for 0a10c34
     AsbjornOlling

     authored
     0a10c34View commit details

     Copy full SHA for 0a10c34

     Browse repository at this point

   • vulkan: add GATED_DELTA_NET op support (#20334)

     Show description for deee238
     

     ProgenyAlpha

     and

     claude

     authored
     deee238View commit details

     Copy full SHA for deee238

     Browse repository at this point

   • convert : better mtp check and fix return [no ci] (#20419)

     CISC

     authored
     c3e3f9eView commit details

     Copy full SHA for c3e3f9e

     Browse repository at this point

   • vulkan: fix SSM_CONV PP scaling with large ubatch sizes (#20379)

     Show description for 40c550d
     

     ProgenyAlpha

     and

     claude

     authored
     40c550dView commit details

     Copy full SHA for 40c550d

     Browse repository at this point

   • New conversations now auto-select the first loaded model (#20403)

     Show description for de19015
     ServeurpersoCom

     authored
     de19015View commit details

     Copy full SHA for de19015

     Browse repository at this point

   • ggml-virtgpu: Fix some build commands (#20341)

     yomaytk

     authored
     0503996View commit details

     Copy full SHA for 0503996

     Browse repository at this point

   • metal : avoid divisions in bin kernel (#20426)

     Show description for e4cff09
     ggerganov

     authored
     e4cff09View commit details

     Copy full SHA for e4cff09

     Browse repository at this point

   • ci: Setup self-hosted CI for Intel Linux Vulkan backend (#20154)

     rillomas

     authored
     4cc6eb1View commit details

     Copy full SHA for 4cc6eb1

     Browse repository at this point

   • vulkan: fix l2_norm epsilon handling (#20350)

     jeffbolznv

     authored
     246ffc4View commit details

     Copy full SHA for 246ffc4

     Browse repository at this point

   • vulkan: fix OOB check in flash_attn_mask_opt (#20296)

     jeffbolznv

     authored
     aa429cfView commit details

     Copy full SHA for aa429cf

     Browse repository at this point

   • vulkan: Fix ErrorOutOfHostMemory on Intel GPU when loading large models with --no-mmap (#20059)

     Show description for 5866e3b
     rillomas

     authored
     5866e3bView commit details

     Copy full SHA for 5866e3b

     Browse repository at this point

   • opencl: use larger workgroup size for get_rows (#20316)

     lhez

     authored
     0516e04View commit details

     Copy full SHA for 0516e04

     Browse repository at this point

   • opencl: add cumsum op (#18981)

     Show description for 3d9ab22
     

     shaofeiqi

     and

     lhez

     authored
     3d9ab22View commit details

     Copy full SHA for 3d9ab22

     Browse repository at this point

   • hip: compile debug builds with -O2 on hip to avoid a compiler bug (#20392)

     IMbackK

     authored
     d63aa39View commit details

     Copy full SHA for d63aa39

     Browse repository at this point

   • common/parser: add GigaChatV3/3.1 models support (#19931)

     Show description for a8304b4
     

     Mishusha

     and

     Mishusha

     authored
     a8304b4View commit details

     Copy full SHA for a8304b4

     Browse repository at this point
3. Commits on Mar 11, 2026

   • model : add support for Phi4ForCausalLMV (#20168)

     Show description for fdb1764
     

     dranger003

     and

     ngxson

     authored
     fdb1764View commit details

     Copy full SHA for fdb1764

     Browse repository at this point

   • graph : add optional scale parameter to build_lora_mm [no ci] (#20427)

     richarddd

     authored
     1eea6a2View commit details

     Copy full SHA for 1eea6a2

     Browse repository at this point

   • common : fix --n-cpu-moe, --cpu-moe for models with fused gate + up (#20416)

     ddh0

     authored
     4a748b8View commit details

     Copy full SHA for 4a748b8

     Browse repository at this point

   • ggml-webgpu: Add supports for `GGML_OP_REPEAT` (#20230)

     Show description for f2ab047
     yomaytk

     authored
     f2ab047View commit details

     Copy full SHA for f2ab047

     Browse repository at this point

   • llama : enable chunked fused GDN path (#20340)

     Show description for d28961d
     

     6 peopleauthored
     d28961dView commit details

     Copy full SHA for d28961d

     Browse repository at this point

   • llama : whitespace cleanup (#20422)

     CISC

     authored
     f90bd1dView commit details

     Copy full SHA for f90bd1d

     Browse repository at this point

   • ggml : add NVFP4 quantization type support (#19769)

     Show description for 5eae9cb
     

     richarddd

     and

     CISC

     authored
     5eae9cbView commit details

     Copy full SHA for 5eae9cb

     Browse repository at this point

   • benches : add nemotron super (#20420)

     ggerganov

     authored
     3ca19b0View commit details

     Copy full SHA for 3ca19b0

     Browse repository at this point