# Zhipu AI Releases Open-Source GLM-5.1 MoE Model

> The new bilingual model from the Chinese AI firm features an efficient Mixture-of-Experts architecture and a fully permissive MIT license.

Published by The Open Weights on Apr 3, 2026. Canonical: https://theopenweights.com/news/glm-5-1-ufei

## Key facts

- Company: Zhipu AI
- Model: GLM-5.3
- Version: 5.1
- Category: Text / LLM
- Modalities: Text / LLM, Reasoning
- License: MIT (Open weights, commercial use allowed)
- Context window: 203K tokens
- Architecture: GlmMoeDsaForCausalLM
- Precision: BF16
- Weights size: 1507.7 GB
- Significance: major
- Published: 2026-04-03
- Last verified: 2026-06-17
- Hugging Face: https://huggingface.co/zai-org/GLM-5.1
- Canonical URL: https://theopenweights.com/news/glm-5-1-ufei

Chinese AI research firm Zhipu AI has released the weights for GLM-5.1, a new large language model designed for strong performance in both English and Chinese. In a significant move for the open-source community, the model is available under the highly permissive MIT license, allowing for broad commercial and research use.

## An Efficient Architecture

GLM-5.1 is built using a Mixture-of-Experts (MoE) architecture, a technique that improves computational efficiency by only activating a fraction of the model's total parameters for any given input. This approach, popularized by models like Mistral's Mixtral 8x7B, can lead to faster inference times and lower resource requirements compared to dense models of a similar size.

The model also incorporates Deep State-Space Attention (DSA), a more recent alternative to the standard attention mechanism in transformers. While the release card on the [Hugging Face Hub](https://huggingface.co/zai-org/GLM-5.1) does not specify a context length or total parameter count, this architecture suggests a focus on scalable and efficient processing.

This release adds another powerful, commercially-friendly MoE model to the open-source landscape. Its strong bilingual capabilities make it a compelling option for developers building applications that need to serve both Chinese and English-speaking users, a domain where high-quality open models have been less common.

## Get the model

- [Hugging Face](https://huggingface.co/zai-org/GLM-5.1)

## Sources

- [zai-org/GLM-5.1](https://huggingface.co/zai-org/GLM-5.1) — Hugging Face, Apr 3, 2026

---
Source: The Open Weights (https://theopenweights.com/). Aggregated and written by Claude, curated by humans. Cite as: "Zhipu AI Releases Open-Source GLM-5.1 MoE Model", The Open Weights, Apr 3, 2026, https://theopenweights.com/news/glm-5-1-ufei