# Hugging Face Transformers v5.10.1 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v5.10.1
- **Released:** Jun 03, 2026
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v5.10.1/

## Executive Summary (TL;DR)
Hugging Face Transformers v5.10.1. Source highlight: v5.10.0 was yanked as we publish on a corrupted branch. Sorry everyone, this happens when we rush a release!!!

## Breaking Changes & Deprecations
- Breaking changes The Gemma4 vision pooler now casts inputs to float32 before scaling to prevent float16 overflow (inf saturation) with large checkpoints, which may cause minor numerical differences in outputs for users running Gemma-4 vision models in float16.

## New Features
- Add Sapiens2 Model (#45919) by @guarin in #45919
- Add Deepseek-OCR-2 model (#45075) by @thisisiron in #45075
- feat: Add support for JetBrains' `Mellum` v2 code generation model (#46112) by @shadeMe in #46112
- 🚨 [ALM] Add base model without head (#45534)
- Add Deepseek-OCR-2 model (#45075)
- Add Expectations for pipeline token classification tests (#46151)

## Bug Fixes
- **No Vision Tower**: Raw pixel patches are projected directly into LM space via a `Dense + LayerNorm` pipeline with factorized 2D positional embeddings, replacing the vision encoder.
- **No Audio Tower**: Raw 16 kHz waveform samples are chunked into fixed-length frames and projected through a simple `RMSNorm → Linear` pipeline, replacing the mel spectrogram + Conformer encoder.
- 🚨 Fix float16 overflow in Gemma4 vision pooler (#46277)
- Fix dsv4 dequant + tp/ep (#46378) by @IlyasMoutawwakil in [#46378]
- fix series of bugs for model parallel beam search (#46280) by @kaixuanliu in [#46280]
- Fix model parallel issue for altclip model and ChineseClip model (#45487) by @kaixuanliu in [#45487]

## Upgrade Command
```bash
python -m pip install 'transformers==5.10.1'
```