v1.2.1

What's Changed

• fix Z-Image fp16_accumulation argument incompatibility by @devgdovg in #781
• docs: update icon file for faviicon by @ttio2tech in #789
• fix: crash by illegal memory access by @devgdovg in #782
• release: v1.2.1 by @lmxyy in #792

Full Changelog: v1.2.0...v1.2.1

What's Changed

• fix Z-Image fp16_accumulation argument incompatibility by @devgdovg in #781
• docs: update icon file for faviicon by @ttio2tech in #789
• fix: crash by illegal memory access by @devgdovg in #782
• release: v1.2.1 by @lmxyy in #792

Full Changelog: v1.2.0...v1.2.1

v1.2.0

What's Changed

• docs: fix typo in workflows/zimage.rst by @calebowlin in #733
• fuse qkv/norm/rope in z-image attention module by @devgdovg in #766
• chore: adapt new comfyui version for qwen-image models by @devgdovg in #761
• feat: Nunchaku Z-Image model to load and run LoRAs with native LoRALoader node provided by ComfyUI by @devgdovg in #759
• feat: for z-image fp16 (Turing GPU) support by @devgdovg in #758
• chore: release v1.2.0 by @lmxyy in #771

New Contributors

• @calebowlin made their first contribution in #733

Full Changelog: v1.1.0...v1.2.0

v1.1.0

What's Changed

• chore: update the publish CI by @lmxyy in #685
• fix: fix lora loader for latest ComfyUI by @devgdovg in #714
• chore: release v1.1.0 by @lmxyy in #740

Full Changelog: v1.0.2...v1.1.0

ComfyUI-nunchaku v1.0.2

What's Changed

• chore: update the pulish ci to create a release by @lmxyy in #588
• feat: upload the workflow for Qwen-Image-Edit-2509 by @lmxyy in #601
• chore: add new testings for nunchaku comfyui workflows by @ttio2tech in #604
• chore: switch the test backend to hiddenswitch's ComfyUI LTS by @lmxyy in #634
• chore: update the workflow for the pr test by @lmxyy in #635
• chore: fix the test ci by @lmxyy in #637
• docs: update the docs for the new tests by @lmxyy in #636
• fix: better nunchaku version check by @lmxyy in #640
• feat: use centralized version file + installer node edit by @Juste-Leo2 in #587
• Update install_wheel.json to install the latest stable version. by @Joshua-Chin in #602
• feat: add the Script to Build ComfyUI with Nunchaku Built In by @kungtalon in #586
• chore: fix the desktop exe building issue by @lmxyy in #649
• feat: new tests for nunchaku comfyui from lmxyy & ttio2tech, tested and verified on comfyui 0.3.63 upstream and 0.3.60 LTS (#641) by @lmxyy in #650
• fix: QwenImage with RES4LYF sampler by @devgdovg in #600
• chore: merge the pulid dependency by @lmxyy in #661
• docs: update some docs by @lmxyy in #662
• chore: Add installation guide of Nunchaku Packages ComfyUI by @kungtalon in #669
• chore: wrap up v1.0.2 by @lmxyy in #682
• chore: release v1.0.2 by @lmxyy in #683
• dosc: amend the docs for v1.0.2 by @lmxyy in #684

New Contributors

• @ttio2tech made their first contribution in #604
• @Joshua-Chin made their first contribution in #602
• @kungtalon made their first contribution in #586
• @devgdovg made their first contribution in #600

Full Changelog: v1.0.1...v1.0.2

ComfyUI-nunchaku v1.0.1

What's Changed

• fix: fix the ControlNet of Qwen-Image by @wallen0322 in #537
• fix: changed device arg to meta for T5XXLModel ctor by @nathandoak in #525
• fix: Ensure node initializes correctly without internet by @Juste-Leo2 in #543
• feat: qwen-image-edit support by @lmxyy in #552
• feat: expose num_blocks_on_gpu and use_pin_memory by @lmxyy in #570
• feat: Enable offline init and improve Nunchaku installer node by @Juste-Leo2 in #564
• chore: release v1.0.1 by @lmxyy in #573

New Contributors

• @wallen0322 made their first contribution in #537
• @nathandoak made their first contribution in #525

Full Changelog: v1.0.0...v1.0.1

ComfyUI-nunchaku v1.0.0

What's Changed

• chore: fix the wheel source link by @lmxyy in #453
• feat: Qwen-Image support by @lmxyy in #463
• chore: release v1.0.0dev1, support QwenImage by @lmxyy in #467
• feat: cache mechanism file separation by @Bluear7878 in #474
• feat: Nunchaku Installer Node Enhancements by @Juste-Leo2 in #481
• fix(nunchaku-installer): add ModelScope API as version source with timeout by @Juste-Leo2 in #493
• feat: add NunchakuFluxLoraStack node for multi-LoRA support by @Roywhite in #489
• fix: close the NVFP4 performance gap between the Python backend and C backend by @lmxyy in #498
• feat: support CPU offloading for Qwen-Image to reduce the VRAM usage by @lmxyy in #520
• fix: fix the bug when disabling cpu_offloading by @lmxyy in #521
• chore: release v1.0.0 by @lmxyy in #523
• chore: officially release v1.0.0 by @lmxyy in #524

New Contributors

• @Juste-Leo2 made their first contribution in #481
• @Roywhite made their first contribution in #489

Full Changelog: v0.3.4...v1.0.0

ComfyUI-nunchaku v0.3.4

What's Changed

• Update README_ZH.md by @PaoFanAIGC in #316
• docs: update the wechat link by @lmxyy in #395
• docs: initialize the docs by @lmxyy in #394
• Merge dev into main by @lmxyy in #397
• chore: update a workflow to build docs by @lmxyy in #405
• chore: better test ci by @lmxyy in #415
• feat: add IP-Adapter support to ComfyUI by @Bluear7878 in #305
• feat: Optimize NunchakuFluxPuLIDApplyV2 processing to support batch image embedding by @Yanick112 in #298
• fix: comma character encoding and proper offload action gerund form by @Ristovski in #422
• fix: bring block cache back by @senlyu163 in #428
• feat: update doc for latest huggingface cli usage by @BoynChan in #431
• fix: Improve compatibility with nunchaku versions by @senlyu163 in #434
• docs: fix the broken official Nunchaku project link (404) in the README. by @gaowei-space in #440
• chore: release v0.3.3 by @lmxyy in #446
• chore: change the version to v0.3.4 by @lmxyy in #447

New Contributors

• @Bluear7878 made their first contribution in #305
• @Yanick112 made their first contribution in #298
• @Ristovski made their first contribution in #422
• @senlyu163 made their first contribution in #428
• @BoynChan made their first contribution in #431
• @gaowei-space made their first contribution in #440

Full Changelog: v0.3.3...v0.3.4

ComfyUI-nunchaku v0.3.3

What's Changed

• feat: support flux.1-kontext by @lmxyy in #314
• chore: release v0.3.3 by @lmxyy in #315

Full Changelog: v0.3.2...v0.3.3

ComfyUI-nunchaku v0.3.2

What's Changed

• fix: installer node can be used when nunchaku is not installed by @lmxyy in #271
• chore: add tests for the koyha LoRA by @lmxyy in #285
• feat: reusing PuLID folders and add the option of start and end by @lmxyy in #291
• chore: release v0.3.2 by @lmxyy in #293

Full Changelog: v0.3.1...v0.3.2

ComfyUI-nunchaku v0.3.1

What's Changed

• docs: update README_ZH.md by @PaoFanAIGC in #242
• fix: fix the NunchakuT5Encoder on lowvram mode by @lmxyy in #256
• fix: bring the cache back by @lmxyy in #262
• feat: catch the import errors of the optional nodes by @lmxyy in #263
• feat: skip loading LoRA when LoRA weight is 0 by @lmxyy in #264
• feat: add a wheel installation node by @lmxyy in #269
• chore: release v0.3.1 by @lmxyy in #270

Full Changelog: v0.3.0...v0.3.1