# Hugging Face Transformers v5.19.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v5.19.0
- **Released:** Oct 06, 2026
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v5.19.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v5.19.0. Source highlight: <img width="2716" height="2308" alt="image" src="https://github.com/user-attachments/assets/84734e74-163d-4d12-b166-ffcf6749d563" />

## Breaking Changes & Deprecations
- Breaking changes All MoE models whose routers compute logits now return them when `output_router_logits=True`, following the Qwen3-MoE pattern (a `router_logits` recorder on the base model, `MoeModelOutputWithPast` from the backbone, and a MoE causal LM output from the head), so code that relied on the previous outputs or their absence should read the router logits from these output classes.

## New Features
- [`GTE`] Add memory mixin (#49234)
- [CB] Make CB more device agnostic and add support for XPU (#49156)

## Bug Fixes
- 🚨 [CB] 🚨 Little fixes before removing "paged" (#49069)
- [`distributed`] Add expert-parallel token dispatch, default for Qwen3 MoE (#48865) by @3outeille in [#48865]
- Fix pp chart2table inference (#49225) by @zucchini-nlp in [#49225]
- Fix quantized cache (#48700) by @jiqing-feng in [#48700]
- Revert "[CI] Temporarily disable AMD scheduled CI trigger" (#49271) (#49362) by @ydshieh in [#49362]
- ALM testx fIxing... (#49355) by @zucchini-nlp in [#49355]

## Upgrade Command
```bash
python -m pip install 'transformers==5.19.0'
```