# Hugging Face Transformers v5.0.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v5.0.0
- **Released:** Jan 26, 2026
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v5.0.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v5.0.0. Source highlight: Transformers v5 release notes

## Breaking Changes & Deprecations
- Breaking Changes * **`report_to` now defaults to `"none"`**
- Biggest breaking change is in `transformers chat`. This command starts a terminal UI to interact with a chat model.
- Revert some breaking changes bnb  by @SunMarc in #41581
- [Trainer] [Breaking change] `use_cache` default to `False`  by @SunMarc in #41585
- added more breaking changes * [cleanup] Don't use Repository in create_dummy_models.py script (#42380) * [cleanup] Remove deprecated load config from file (#42383)

## New Features
- `added_tokens.json` - added tokens are now stored in `tokenizer.json`.
- `SpecialTokensMixin`: Merged into `PreTrainedTokenizerBase` to simplify the tokenizer architecture.
- `special_tokens_map`: Now only stores named special token attributes (e.g., `bos_token`, `eos_token`). Use `extra_special_tokens` for additional special tokens (formerly `additional_special_tokens`). `all_special_tokens` includes both named and extra tokens.
- Add buffers to _init_weights for ALL models
- No more head masking, see #41076. This feature allowed to turn off certain heads during the attention calculation and only worked for eager.
- No more relative positional biases in Bert-like models, see #41170. This feature was introduced to allow relative position scores within attention calculations (similar to T5). However, this feature is barely used in official models and a lot of complexity instead. It also only worked with eager.

## Improvements
- [saving] Simplify general logic
- Do not rely on config for inferring model dtype
- Improve BatchFeature: stack list and lists of torch tensors
- Remove tied weights from internal attribute if they are not tied
- Simplify tie weights logic
- [loading] Really initialize on meta device for huge perf gains

## Bug Fixes
- Bugfixes and improvements
- Enforce call to post_init and fix all of them
- `tpu_metrics_debug` -> `debug`
- `JetMoe` Fix jetmoe after #40132  by @ArthurZucker in #41324
- Fixed tiny incorrect import in `gemma3`  by @Sai-Suraj-27 in #41354
- Rope for Qwen2--5-vl  by @zucchini-nlp in #41173

## Upgrade Command
```bash
python -m pip install 'transformers==5.0.0'
```