# Hugging Face Transformers v5.9.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v5.9.0
- **Released:** May 20, 2026
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v5.9.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v5.9.0. Source highlight: Command A+ is a Mixture-of-Experts (MoE) language model from Cohere that features a hybrid attention pattern combining sliding window and full attention layers. The model incorporates both shared and routed experts and supports a very large context window for processing extensive text sequences.

## Breaking Changes & Deprecations
- Breaking changes The `text_embeds` input for SAM3, EdgeTAM, and SAM3-Lite-Text models now expects full text embeddings instead of just pooler outputs, aligning with other models in the library — users must update their inputs accordingly.

## New Features
- Add new cohere2_moe model (#46115) by @Cyrilvallez in #46115
- Add hrm text (#46025) by @abcd1927 in #46025
- [MultimodalLM] add language_model to the get/set_input_embeddings logic (#46029)
- [CB] [Major] Add tensor paralellism (#45821)
- Add hrm text (#46025)

## Bug Fixes
- 🚨Fix memory leaks caused by lru decorators in vision models (#45922)
- Fix Gemma4 generation from inputs_embeds and per_layer_inputs (#46049) by @Cyrilvallez in [#46049]
- Fix AttributeError in RAG generate() for missing config fields (#46035) by @Sriniketh24 in [#46035]
- Remove mask visualization tool from `masking_utils.py` (#46066) by @Cyrilvallez in [#46066]
- fix: owned_by field in GET /v1/models returns list instead of string (#46006) by @nileshpatil6 in [#46006]
- [CB] Remove OpenTelemetry (#45984) by @remi-or in [#45984]

## Upgrade Command
```bash
python -m pip install 'transformers==5.9.0'
```