v4.56.1 is a patch release. Key changes include:

• Fix broken Llama4 accuracy in MoE part;
• fix pipeline dtype.

v4.56.0-Embedding-Gemma-preview is a pre-release release. Key changes include:

• A new model is added to transformers: Embedding Gemma It is added on top of the v4.56.0 release, and can be installed from the following tag: v4.56.0-Embedding…;
• In order to install this version, please install with the following command:.

v4.56.0 is a minor release. Key changes include:

• Add Dino v3 by @qubvel in #40167;
• Music continuation : Better modeling of musical semantics yields more coherent continuations..

v4.55.4 is a patch release. Key changes include:

• Patch v4.55.4 There was a mick mack on our side when cherry-picking the commit #40197 which led to a wrong commit in the patch!;
• This patch is just the official fix for #40197!.

v4.55.3 is a patch release. Key changes include:

• FlashAttention-2 / Ascend NPU – Fix “unavailable” runtime error (#40151) by @FightingZhen;
• FlashAttention kwargs – Revert FA kwargs preparation to resolve regression (#40161) by @Cyrilvallez.

v4.55.2 is a patch release. Key changes include:

• Patch release 4.55.2!;
• only affects FA2 generations!.

v4.55.1 is a patch release. Key changes include:

• Idefics2, Idefics3, SmolVLM – Fix tensor device issue (#39975) by @qgallouedec;
• Merge conflicts – Fix merge conflicts from previous changes by @vasqu.

4.55.0-GLM-4.5V-preview is a pre-release release. Key changes include:

• Image reasoning (scene understanding, complex multi-image analysis, spatial recognition);
• Video understanding (long video segmentation and event recognition).

v4.55.0 is a minor release. Key changes include:

• 21B and 117B total parameters, with 3.6B and 5.1B active parameters, respectively.;
• 4-bit quantization scheme using mxfp4 format. Only applied on the MoE weights. As stated, the 120B fits in a single 80 GB GPU and the 20B fits in a single 16GB….

4.54.1 is a patch release. Key changes include:

• Fix Cache.max_cache_len max value for Hybrid models, @manueldeprada, @Cyrilvallez, #39737;
• [modenbert] fix regression, @zucchini-nlp, #39750.

v4.54.0 is a minor release. Key changes include:

• [Ernie 4.5] Add ernie text models by @vasqu in #39228;
• 3B: mistralai/Voxtral-Mini-3B-2507.

v4.53.2-Ernie-4.5-preview is a pre-release release. Key changes include:

• Two new models are added to transformers: Ernie 4.5, and its MoE variant, Ernie 4.5 MoE.;
• In order to install this version, please install with the following command:.

v4.53.3 is a patch release. Key changes include:

• Small path release 4.53.3!;
• ** refactor: remove set_tracer_provider and set_meter_provider calls (https://github.com/huggingface/transformers/pull/39422) from @McPatate.

v4.53.2-modernbert-decoder-preview is a pre-release release. Key changes include:

• A new model is added to transformers: ModernBERT Decoder It is added on top of the v4.53.2 release, and can be installed from the following tag: v4.53.2-modern…;
• In order to install this version, please install with the following command:.

v4.53.2 is a patch release. Key changes include:

• Fix some bug for finetune and batch infer For GLM-4.1V;
• [bugfix] fix flash attention 2 unavailable error on Ascend NPU.

v4.53.1 is a patch release. Key changes include:

• Fix: unprotected import of tp plugin;
• Fix key mapping for VLMs.

v4.53.0 is a minor release. Key changes include:

• Add Dia model by @buttercrab in #38405;
• kyutai/stt-1b-en_fr: a 1B-parameter model capable of transcribing both English and French.

v4.52.4-Kyutai-STT-preview is a pre-release release. Key changes include:

• kyutai/stt-1b-en_fr: a 1B-parameter model capable of transcribing both English and French;
• kyutai/stt-2.6b-en: a 2.6B-parameter model focused solely on English, optimized for maximum transcription accuracy.

v4.52.4-VJEPA-2-preview is a pre-release release. Key changes include:

• A new model is added to transformers: V-JEPA 2 It is added on top of the v4.52.4 release, and can be installed from the following tag: v4.52.4-VJEPA-2-preview.;
• In order to install this version, please install with the following command:.

v4.52.4-ColQwen2-preview is a pre-release release. Key changes include:

• A new model is added to transformers: ColQwen2 It is added on top of the v4.52.4 release, and can be installed from the following tag: v4.52.4-ColQwen2-preview.;
• In order to install this version, please install with the following command:.

v4.52.4 is a patch release. Key changes include:

• [qwen-vl] Look for vocab size in text config;
• Fix convert to original state dict for VLMs.

v4.52.3 is a patch release. Key changes include:

• Fix tp error when torch distributed is already initialized (#38294) by @SunMarc;
• Protect ParallelInterface (#38262) by @ArthurZucker and @LysandreJik.

v4.52.2 is a patch release. Key changes include:

• Clearer error on import failure (#38257) by @LysandreJik;
• Verified tp plan should not be NONE (#38255) by @NouamaneTazi and @ArthurZucker.

v4.52.1 is a patch release. Key changes include:

• SAM-HQ produces higher quality masks than the original SAM model, particularly for objects with intricate structures and fine details;
• The model predicts binary masks with more accurate boundaries and better handling of thin structures.

v4.51.3-CSM-preview is a pre-release release. Key changes include:

• A new model is added to transformers: CSM It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-CSM-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-GraniteMoeHybrid-preview is a pre-release release. Key changes include:

• A new model is added to transformers: GraniteMoeHybrid It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-GraniteM…;
• In order to install this version, please install with the following command:.

v4.51.3-D-FINE-preview is a pre-release release. Key changes include:

• A new model is added to transformers: D-FINE It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-D-FINE-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-SAM-HQ-preview is a pre-release release. Key changes include:

• SAM-HQ produces higher quality masks than the original SAM model, particularly for objects with intricate structures and fine details;
• The model predicts binary masks with more accurate boundaries and better handling of thin structures.

v4.51.3-BitNet-preview is a pre-release release. Key changes include:

• A new model is added to transformers: BitNet It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-BitNet-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-LlamaGuard-preview is a pre-release release. Key changes include:

• A new model is added to transformers: LlamaGuard It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-LlamaGuard-pre…;
• In order to install this version, please install with the following command:.

v4.56.1 is a patch release. Key changes include:

• Fix broken Llama4 accuracy in MoE part;
• fix pipeline dtype.

v4.56.0-Embedding-Gemma-preview is a pre-release release. Key changes include:

• A new model is added to transformers: Embedding Gemma It is added on top of the v4.56.0 release, and can be installed from the following tag: v4.56.0-Embedding…;
• In order to install this version, please install with the following command:.

v4.56.0 is a minor release. Key changes include:

• Add Dino v3 by @qubvel in #40167;
• Music continuation : Better modeling of musical semantics yields more coherent continuations..

v4.55.4 is a patch release. Key changes include:

• Patch v4.55.4 There was a mick mack on our side when cherry-picking the commit #40197 which led to a wrong commit in the patch!;
• This patch is just the official fix for #40197!.

v4.55.3 is a patch release. Key changes include:

• FlashAttention-2 / Ascend NPU – Fix “unavailable” runtime error (#40151) by @FightingZhen;
• FlashAttention kwargs – Revert FA kwargs preparation to resolve regression (#40161) by @Cyrilvallez.

v4.55.2 is a patch release. Key changes include:

• Patch release 4.55.2!;
• only affects FA2 generations!.

v4.55.1 is a patch release. Key changes include:

• Idefics2, Idefics3, SmolVLM – Fix tensor device issue (#39975) by @qgallouedec;
• Merge conflicts – Fix merge conflicts from previous changes by @vasqu.

4.55.0-GLM-4.5V-preview is a pre-release release. Key changes include:

• Image reasoning (scene understanding, complex multi-image analysis, spatial recognition);
• Video understanding (long video segmentation and event recognition).

v4.55.0 is a minor release. Key changes include:

• 21B and 117B total parameters, with 3.6B and 5.1B active parameters, respectively.;
• 4-bit quantization scheme using mxfp4 format. Only applied on the MoE weights. As stated, the 120B fits in a single 80 GB GPU and the 20B fits in a single 16GB….

4.54.1 is a patch release. Key changes include:

• Fix Cache.max_cache_len max value for Hybrid models, @manueldeprada, @Cyrilvallez, #39737;
• [modenbert] fix regression, @zucchini-nlp, #39750.

v4.54.0 is a minor release. Key changes include:

• [Ernie 4.5] Add ernie text models by @vasqu in #39228;
• 3B: mistralai/Voxtral-Mini-3B-2507.

v4.53.2-Ernie-4.5-preview is a pre-release release. Key changes include:

• Two new models are added to transformers: Ernie 4.5, and its MoE variant, Ernie 4.5 MoE.;
• In order to install this version, please install with the following command:.

v4.53.3 is a patch release. Key changes include:

• Small path release 4.53.3!;
• ** refactor: remove set_tracer_provider and set_meter_provider calls (https://github.com/huggingface/transformers/pull/39422) from @McPatate.

v4.53.2-modernbert-decoder-preview is a pre-release release. Key changes include:

• A new model is added to transformers: ModernBERT Decoder It is added on top of the v4.53.2 release, and can be installed from the following tag: v4.53.2-modern…;
• In order to install this version, please install with the following command:.

v4.53.2 is a patch release. Key changes include:

• Fix some bug for finetune and batch infer For GLM-4.1V;
• [bugfix] fix flash attention 2 unavailable error on Ascend NPU.

v4.53.1 is a patch release. Key changes include:

• Fix: unprotected import of tp plugin;
• Fix key mapping for VLMs.

v4.53.0 is a minor release. Key changes include:

• Add Dia model by @buttercrab in #38405;
• kyutai/stt-1b-en_fr: a 1B-parameter model capable of transcribing both English and French.

v4.52.4-Kyutai-STT-preview is a pre-release release. Key changes include:

• kyutai/stt-1b-en_fr: a 1B-parameter model capable of transcribing both English and French;
• kyutai/stt-2.6b-en: a 2.6B-parameter model focused solely on English, optimized for maximum transcription accuracy.

v4.52.4-VJEPA-2-preview is a pre-release release. Key changes include:

• A new model is added to transformers: V-JEPA 2 It is added on top of the v4.52.4 release, and can be installed from the following tag: v4.52.4-VJEPA-2-preview.;
• In order to install this version, please install with the following command:.

v4.52.4-ColQwen2-preview is a pre-release release. Key changes include:

• A new model is added to transformers: ColQwen2 It is added on top of the v4.52.4 release, and can be installed from the following tag: v4.52.4-ColQwen2-preview.;
• In order to install this version, please install with the following command:.

v4.52.4 is a patch release. Key changes include:

• [qwen-vl] Look for vocab size in text config;
• Fix convert to original state dict for VLMs.

v4.52.3 is a patch release. Key changes include:

• Fix tp error when torch distributed is already initialized (#38294) by @SunMarc;
• Protect ParallelInterface (#38262) by @ArthurZucker and @LysandreJik.

v4.52.2 is a patch release. Key changes include:

• Clearer error on import failure (#38257) by @LysandreJik;
• Verified tp plan should not be NONE (#38255) by @NouamaneTazi and @ArthurZucker.

v4.52.1 is a patch release. Key changes include:

• SAM-HQ produces higher quality masks than the original SAM model, particularly for objects with intricate structures and fine details;
• The model predicts binary masks with more accurate boundaries and better handling of thin structures.

v4.51.3-CSM-preview is a pre-release release. Key changes include:

• A new model is added to transformers: CSM It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-CSM-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-GraniteMoeHybrid-preview is a pre-release release. Key changes include:

• A new model is added to transformers: GraniteMoeHybrid It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-GraniteM…;
• In order to install this version, please install with the following command:.

v4.51.3-D-FINE-preview is a pre-release release. Key changes include:

• A new model is added to transformers: D-FINE It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-D-FINE-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-SAM-HQ-preview is a pre-release release. Key changes include:

• SAM-HQ produces higher quality masks than the original SAM model, particularly for objects with intricate structures and fine details;
• The model predicts binary masks with more accurate boundaries and better handling of thin structures.

v4.51.3-BitNet-preview is a pre-release release. Key changes include:

• A new model is added to transformers: BitNet It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-BitNet-preview.;
• In order to install this version, please install with the following command:.

v4.51.3-LlamaGuard-preview is a pre-release release. Key changes include:

• A new model is added to transformers: LlamaGuard It is added on top of the v4.51.3 release, and can be installed from the following tag: v4.51.3-LlamaGuard-pre…;
• In order to install this version, please install with the following command:.