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-pipelinespipelinessize/Ltestsutils
  Unread
  huggingface/diffusers#14159·yiyixuxu opened Jul 10, 2026 · Updated Oct 2, 2026
  32

• Allow tensor parallelism and context parallelism in one ParallelConfig

  hooksmodelsno-issue-neededsize/Ltests
  Unread
  huggingface/diffusers#14725·whn09 opened Sep 7, 2026 · Updated Oct 2, 2026
  8

  

• fix: scope save_pretrained shard cleanup to the saved variant (avoid deleting another variant's shards)

  fixes-issuemodelssize/Mtests
  Unread
  huggingface/diffusers#14844·AloysJehwin opened Sep 22, 2026 · Updated Oct 4, 2026
  1

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

  documentationhooksmodelssize/Ltestsutils
  Unread
  huggingface/diffusers#14039·JingyaHuang opened Jun 22, 2026 · Updated Oct 5, 2026
  45

• feat: implement cache_context for pab and fastercache.

  documentationhooksmodelsmodular-pipelinespipelinessize/Ltests
  Unread
  huggingface/diffusers#14900·sayakpaul opened Sep 29, 2026 · Updated Oct 5, 2026
  8

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

  fixes-issuelorasize/Ltestsutils
  Unread
  huggingface/diffusers#14163·christopher5106 opened Jul 10, 2026 · Updated Oct 5, 2026
  21