# Transformers v5.23.0 — LFM3, Inkling 2 and a rebuilt model export

> Seven new model types and a new export API land in Hugging Face's model library.

- Org: Hugging Face
- Date: 2026-11-13
- Category: tool, repo
- Source: https://github.com/huggingface/transformers/releases/tag/v5.23.0
- Canonical page: https://ai-tldr.dev/releases/huggingface-transformers-5-23-0/

## In short

Transformers v5.23.0 adds Liquid AI's LFM3 and LFM3-VL, Thinking Machines' Inkling 2 Small, Dots 3 Note Preview and InternLM3. Exports now return a callable ExportedModel, and expert parallelism gets a selectable implementation.

## What it is

Transformers v5.23.0 adds seven model types: Inkling 2 Small from Thinking Machines Lab, Liquid AI's third-generation LFM3 text models and LFM3-VL, Dots 3 Note Preview (a mixture of experts that takes text, image, video and audio), the IQuestCoder code models, InternLM3 8B and an experimental decision head called JEV. The bigger change is an overhaul of model export.

## How it works

export() now returns an ExportedModel that you can call like the original model, push to the Hub and reload on the backend's runtime, with ONNX Runtime fusions applied at export time; output_path options are gone in favour of save_pretrained. For large mixture-of-experts models, ep_implementation picks the expert-parallel strategy (all-reduce for inference, all-to-all for training), and DistributedConfig gains HSDP.

## Why it matters

Teams that ship exported models get a simpler path from training code to a deployable artifact. The release also has breaking changes to check before upgrading: the ep_router style is removed, the [audio] extra no longer installs librosa or phonemizer, and LM-boosted Wav2Vec2 decoding is deprecated.

## Who it's for

ML engineers using Hugging Face models

## Try it

pip install -U transformers==5.23.0

## Links

- [Release Notes](https://github.com/huggingface/transformers/releases/tag/v5.23.0)
- [Repository](https://github.com/huggingface/transformers)
- [Docs](https://huggingface.co/docs/transformers)