# Hugging Face Transformers v4.53.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v4.53.0
- **Released:** Jun 26, 2025
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v4.53.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v4.53.0. Source highlight: Gemma 3n models are designed for efficient execution on low-resource devices. They are capable of multimodal input, handling text, image, video, and audio input, and generating text outputs, with open weights for pre-trained and instruction-tuned variants. These models were trained with data in over

## Breaking Changes & Deprecations
- Breaking changes Several minimal breaking changes aiming to bring clearer defaults while greatly simplifying the library have been merged.
- Several minimal breaking changes aiming to bring clearer defaults while greatly simplifying the library have been merged.

## New Features
- Add Dia model  by @buttercrab in #38405
- Add kyutai stt  by @eustlb in #38909
- Add V-JEPA 2  by @qubvel in #38746
- Add Arcee model support  by @Crystalcareai in #38621
- Add ColQwen2 to 🤗 transformers  by @tonywu71 in #35778
- Add support for MiniMax's MiniMax-Text-01  by @geetu040 in #35831

## Improvements
- CI reporting improvements (#38230)
- Improve `test_initialization` (#38607)
- Improve `test_initialization` for `SwiftFormer` (#38636)
- Improve typing in TrainingArgument (#36944)

## Bug Fixes
- 🚨🚨 Fix initialization of Mask2Former  by @Cyrilvallez in #38864
- :rotating_light: :rotating_light: Fix custom code saving  by @Rocketknight1 in #37716
- CI reporting improvements  by @ydshieh in #38230
- Revert parallelism temporarily  by @LysandreJik in #38240
- tp plan should not be NONE  by @ArthurZucker in #38255
- [Falcon H1] Fix Typo in Integration Test  by @dhiaEddineRhaiem in #38256

## Upgrade Command
```bash
python -m pip install 'transformers==4.53.0'
```