Release 0.42.0

Issues

• [modular] improve auto offload

  documentationmodular-pipelinessize/Ltestsutils
  Unread
  huggingface/diffusers#14304·yiyixuxu opened Jul 27, 2026 · Updated Oct 2, 2026
  9

• [modular] LTX-2.5: two-stage generation as one pipeline

  documentationmodular-pipelinessize/Ltestsutils
  Unread
  huggingface/diffusers#14612·yiyixuxu opened Aug 27, 2026 · Updated Oct 2, 2026
  8

• [Modular]improve loop block

  documentationmodular-pipelinessize/Ltestsutils
  Unread
  huggingface/diffusers#14159·yiyixuxu opened Jul 10, 2026 · Updated Oct 6, 2026
  33

• Allow tensor parallelism and context parallelism in one ParallelConfig

  hooksmodelsno-issue-neededsize/Ltests
  Unread
  huggingface/diffusers#14725·whn09 opened Sep 7, 2026 · Updated Oct 8, 2026
  9

  

• [TPU] TorchTPU backend integration - eager / torch.compile / tp

  documentationhooksmodelspipelinessize/Ltestsutils
  Unread
  huggingface/diffusers#14039·JingyaHuang opened Jun 22, 2026 · Updated Oct 8, 2026
  67

• feat: implement cache_context for pab and fastercache.

  documentationhooksmodelsmodular-pipelinespipelinessize/Ltests
  Unread
  huggingface/diffusers#14900·sayakpaul opened Sep 29, 2026 · Updated Oct 8, 2026
  15

• [LoRA] add LoKr adapter support (Z-Image, Flux2/Klein)

  fixes-issueGood Example PRlorasize/Ltestsutils
  Unread
  huggingface/diffusers#14163·christopher5106 opened Jul 10, 2026 · Updated Oct 8, 2026
  35

• feat: cp support in qwenimage 2.1

  modelssize/Ltests
  Unread
  huggingface/diffusers#14817·sayakpaul opened Sep 20, 2026 · Updated Oct 7, 2026·Draft
  4

• Per-module alphas of kohya text encoder LoRAs collapse to the most common alpha (alpha_pattern keys stop at the parent module)

  buglorapipelines
  huggingface/diffusers#14970 · Filter by author andre-menutole opened on Oct 6, 2026 · Release 0.42.0
  6