Skip to content

Microsoft Announces Aion Instruct 2.3B Model on Hugging Face

Microsoft released Aion Instruct 2.3B, a 2.3 billion parameter model for edge computing.

Sources in the coveragePerformance and quality statements from a vendor are claims unless the coverage cites independent measurements.

Evidence & verification 4 statements

Specifications, source statements and editorial judgement have different scopes. A citation is not an independent performance test.

ParametersSource statement

2.3 billion

huggingface.co ↗Source observation date not recorded.
LicenseSource statement

MIT

huggingface.co ↗Source observation date not recorded.
Library NamesSource statement

PyTorch, ONNX, GGUF

huggingface.co ↗Source observation date not recorded.
Context WindowSource statement

32K tokens

huggingface.co ↗Source observation date not recorded.

Microsoft announced the release of Aion Instruct 2.3B, a 2.3 billion parameter model for edge computing. It is designed for enterprise-grade language understanding and generation, enabling private and cost-effective AI scenarios without cloud connectivity. The model is available on Hugging Face and will be integrated into Windows OS by the end of January 2027.

Key facts

Fact Detail The source says
Parameters 2.3 billion “This 2.3B parameters base model delivers enterprise-grade language understanding and generation capabilities directly at the edge”
License MIT “License: MIT”
Library Names PyTorch, ONNX, GGUF “Library names: PyTorch, ONNX, GGUF”
Context Window 32K tokens “With context window extending to 32K tokens, it can process multi-page documents, lengthy email threads, and complex technical…”

What happened

Microsoft has released Aion Instruct 2.3B, a large language model with 2.3 billion parameters. This model is designed to deliver enterprise-grade language understanding and generation capabilities directly at the edge, ensuring responsive, private, and cost-effective AI-powered scenarios without the need for constant cloud connectivity. The model supports multiple languages including English, German, Spanish, Portuguese, Italian, French, Chinese, and Japanese. It is available under the MIT license and can be used with PyTorch, ONNX, and GGUF libraries. The model has a context window extending to 32K tokens, allowing it to process multi-page documents, lengthy email threads, and complex technical specifications entirely on-device.

What to weigh

  • The model is available on Hugging Face.
  • It supports multiple languages.
  • The model will be integrated into Windows OS by the end of January 2027.

FAQ

What is the model's license?

The model is available under the MIT license.

Which libraries can be used with the model?

The model can be used with PyTorch, ONNX, and GGUF libraries.

When will the model be integrated into Windows OS?

A further optimized version of Aion Instruct, coupled with Microsoft’s content moderation model, will become the Windows OS SLM by the end of January 2027.


Source: Microsoft

Keep following technology

Continue from here.

Explore more coverage or subscribe to Fathom Layer updates.

Back to the news desk ↗

Occasional verified technology updates. Unsubscribe anytime. See our privacy policy.