Commit history

1. Commits on Sep 29, 2026

   • Fixes for last PR. (#16637)

     comfyanonymous

     authored
     9671d89View commit details

     Copy full SHA for 9671d89

     View code at this pointBrowse repository at this point

   • Support w6a8 quantization format (#16483)

     kijai

     authored
     491bf3bView commit details

     Copy full SHA for 491bf3b

     View code at this pointBrowse repository at this point
2. Commits on Sep 24, 2026

   • feat: ming-image support (#16482)

     kijai

     authored
     3b4c0b0View commit details

     Copy full SHA for 3b4c0b0

     View code at this pointBrowse repository at this point
3. Commits on Sep 15, 2026

   • Aimdo 0.5.5 + Auto-detect and enable --fast-disk when the disk is fast (CORE-440) (#16333)

     rattus128

     authored
     7a0b5eeView commit details

     Copy full SHA for 7a0b5ee

     View code at this pointBrowse repository at this point

   • MiniMax-H3 VAE optimizations (#16187)

     kijai

     authored
     b2e31e8View commit details

     Copy full SHA for b2e31e8

     View code at this pointBrowse repository at this point
4. Commits on Sep 12, 2026

   • Fix linear_input_act to respect _full_precision_mm fallback (#16285)

     chelsealong

     authored
     a20738fView commit details

     Copy full SHA for a20738f

     View code at this pointBrowse repository at this point
5. Commits on Sep 8, 2026

   • Cleaner way of enabling quantized mm on text gen but not on text enc. (#16189)

     comfyanonymous

     authored
     249c5a3View commit details

     Copy full SHA for 249c5a3

     View code at this pointBrowse repository at this point

   • Fix quantized text encoder matmul gating and Gemma4 prefill cache positions (#16185)

     kijai

     authored
     b7ebfd7View commit details

     Copy full SHA for b7ebfd7

     View code at this pointBrowse repository at this point
6. Commits on Sep 7, 2026

   • Disable int8 weight-only quantization on devices without torch._int_mm (fixes MPS crash) (#16130)

     chelsealong

     authored
     313a76fView commit details

     Copy full SHA for 313a76f

     View code at this pointBrowse repository at this point
7. Commits on Aug 14, 2026

   • Speedup Gemma4 text generation (CORE-371) (#15054)

     kijai

     authored
     7fe8a61View commit details

     Copy full SHA for 7fe8a61

     View code at this pointBrowse repository at this point
8. Commits on Aug 13, 2026

   • Implement Minimax Music 3 + Core Support for Cuda Graphs (#15570)

     rattus128

     authored
     efd4e95View commit details

     Copy full SHA for efd4e95

     View code at this pointBrowse repository at this point
9. Commits on Aug 9, 2026

   • Make a context manager for cast_bias_weight and use it. (#14750)

     blepping

     authored
     cbbc9daView commit details

     Copy full SHA for cbbc9da

     View code at this pointBrowse repository at this point
10. Commits on Aug 7, 2026

    • Support asym w4a8_int (#15308)

      Show description for 344b439
      kijai

      authored
      344b439View commit details

      Copy full SHA for 344b439

      View code at this pointBrowse repository at this point
11. Commits on Aug 4, 2026

    • restore SDPA non-cudnn small attention bypass (#15296)

      Show description for 40dbdc1
      rattus128

      authored
      40dbdc1View commit details

      Copy full SHA for 40dbdc1

      View code at this pointBrowse repository at this point
12. Commits on Aug 3, 2026

    • feat: Support MiniMax-H3 (CORE-375) (#15224)

      kijai

      authored
      57500fcView commit details

      Copy full SHA for 57500fc

      View code at this pointBrowse repository at this point
13. Commits on Jul 31, 2026

    • Expand k, v when attention backend would fall back to math because gqa. (#15190)

      comfyanonymous

      authored
      a1c4219View commit details

      Copy full SHA for a1c4219

      View code at this pointBrowse repository at this point

    • ops: apply the custom placeholder logic to Linux too (#15181)

      Show description for 7c80628
      rattus128

      authored
      7c80628View commit details

      Copy full SHA for 7c80628

      View code at this pointBrowse repository at this point
14. Commits on Jul 30, 2026

    • Fallback to cudnn attention on linux if flash attention doesn't work. (#15146)

      comfyanonymous

      authored
      f73e8cdView commit details

      Copy full SHA for f73e8cd

      View code at this pointBrowse repository at this point
15. Commits on Jul 28, 2026

    • Load weights to process RAM with MRU policy using pinning infrastructure (#15027)

      rattus128

      authored
      c011755View commit details

      Copy full SHA for c011755

      View code at this pointBrowse repository at this point

    • Support int8 convrot embedding lookup (#15035)

      kijai

      authored
      3d41e3eView commit details

      Copy full SHA for 3d41e3e

      View code at this pointBrowse repository at this point
16. Commits on Jul 10, 2026

    • Fix FP8 activation quantization for >2D activations in mixed_precision_ops (#14643)

      Show description for 62e025a
      liminfei-amd

      authored
      62e025aView commit details

      Copy full SHA for 62e025a

      View code at this pointBrowse repository at this point
17. Commits on Jul 9, 2026

    • Support convrot int4 models. (#14859)

      Show description for 73e84d5
      comfyanonymous

      authored
      73e84d5View commit details

      Copy full SHA for 73e84d5

      View code at this pointBrowse repository at this point
18. Commits on Jul 8, 2026

    • Fix crash when using UNetSelfAttentionMultiply (#14823)

      comfyanonymous

      authored
      ffbecffView commit details

      Copy full SHA for ffbecff

      View code at this pointBrowse repository at this point
19. Commits on Jun 29, 2026

    • Fix int8 mm being skipped on offloaded lora weights. (#14669)

      comfyanonymous

      authored
      79c555cView commit details

      Copy full SHA for 79c555c

      View code at this pointBrowse repository at this point
20. Commits on Jun 26, 2026

    • Fix int8 loras causing lower quality requant with wrong settings. (#14650)

      Show description for 470ac36
      comfyanonymous

      authored
      470ac36View commit details

      Copy full SHA for 470ac36

      View code at this pointBrowse repository at this point
21. Commits on Jun 25, 2026

    • Support int8 models. (#14636)

      comfyanonymous

      authored
      1a510f0View commit details

      Copy full SHA for 1a510f0

      View code at this pointBrowse repository at this point
22. Commits on Jun 12, 2026

    • add --high-ram option (#14437)

      Show description for d7a5527
      rattus128

      authored
      d7a5527View commit details

      Copy full SHA for d7a5527

      View code at this pointBrowse repository at this point
23. Commits on Jun 11, 2026

    • ops: tolerate already force casted dynamic weight (#14410)

      Show description for bda19b2
      rattus128

      authored
      bda19b2View commit details

      Copy full SHA for bda19b2

      View code at this pointBrowse repository at this point
24. Commits on May 30, 2026

    • Threaded Loader performance fixes / improvements (+ Aimdo 0.4.6) (#14116)

      Show description for e154da8
      rattus128

      authored
      e154da8View commit details

      Copy full SHA for e154da8

      View code at this pointBrowse repository at this point
25. Commits on May 26, 2026

    • feat: Microsoft Lens support (CORE-248) (#14077)

      kijai

      authored
      41812faView commit details

      Copy full SHA for 41812fa

      View code at this pointBrowse repository at this point
26. Commits on May 21, 2026

    • Multi-threaded load of models from disk (big load time speedups & Offload to disk) (CORE-43,CORE-152,CORE-164,CORE-165,CORE-117) (#13802)

      Show description for 5aa5ccc
      rattus128

      authored
      5aa5cccView commit details

      Copy full SHA for 5aa5ccc

      View code at this pointBrowse repository at this point
27. Commits on May 19, 2026

    • Fix typo in ops.py (#11925)

      NAM-MAN

      authored
      626b082View commit details

      Copy full SHA for 626b082

      View code at this pointBrowse repository at this point
28. Commits on May 17, 2026

    • Log which quant ops are enabled/emulated. (#13946)

      comfyanonymous

      authored
      f48d2a0View commit details

      Copy full SHA for f48d2a0

      View code at this pointBrowse repository at this point
29. Commits on May 12, 2026

    • feat: Support HiDream-O1-Image (CORE-187) (#13817)

      Show description for 8e53f00
      kijai

      authored
      8e53f00View commit details

      Copy full SHA for 8e53f00

      View code at this pointBrowse repository at this point
30. Commits on May 8, 2026

    • Add support for BiRefNet background remove model (CORE-46) (#12747)

      yousef-rafat

      authored
      d3c18c1View commit details

      Copy full SHA for d3c18c1

      View code at this pointBrowse repository at this point