Now in Foundry: Qwen3-Coder-Next, Qwen3-ASR-1.7B, Z-Image

2026-02-09 · Source: Microsoft Foundry Blog articles · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Data Science & Analytics · Depth: Advanced, medium

Summary

Microsoft Foundry has integrated three new enterprise-grade AI models from Hugging Face, now available for deployment as of February 9, 2026. These include Qwen3-Coder-Next, an 80B MoE model that activates only 3B parameters for efficient coding agent capabilities with a 256k context length. Also featured is Qwen3-ASR-1.7B, a 1.7B parameter model offering state-of-the-art multilingual automatic speech recognition across 52 languages and dialects, outperforming GPT-4o, Gemini-2.5, and Whisper-large-v3 on benchmarks. Finally, Z-Image from Tongyi-MAI is a 6B parameter undistilled text-to-image foundation model supporting full Classifier-Free Guidance for high output diversity and aesthetic versatility, generating images up to 2048x2048 pixels. These models balance performance with practical deployment considerations, making them suitable for production systems.

Key takeaway

For NLP Engineers and Computer Vision Engineers evaluating new foundation models for production, consider integrating Qwen3-Coder-Next for efficient coding agents, Qwen3-ASR-1.7B for superior multilingual speech recognition, or Z-Image for high-quality, diverse text-to-image generation. These models offer competitive performance and practical deployment via Microsoft Foundry, potentially reducing infrastructure costs and simplifying multilingual pipelines.

Key insights

New Hugging Face models offer efficient, high-performance AI for coding, multilingual ASR, and advanced image generation.

Principles

• Efficiency via sparse activation (MoE)
• Multilingual capability in single model
• Undistilled models preserve quality

Method

Microsoft Foundry enables direct deployment of Hugging Face models to managed endpoints via its model catalog or one-click deployment from the Hugging Face Hub, configuring secure, scalable inference.

In practice

• Deploy Qwen3-Coder-Next for local coding agents
• Use Qwen3-ASR-1.7B for multilingual call center analytics
• Generate diverse images with Z-Image for e-commerce

Topics

• Qwen3-Coder-Next
• Multilingual ASR
• Z-Image Text-to-Image
• AI Coding Agents
• Microsoft Foundry Deployment

Code references

• QwenLM/Qwen3-Coder

Best for: NLP Engineer, Computer Vision Engineer, Machine Learning Engineer, AI Engineer, Data Scientist

Related on AIssential

• Qwen3-Coder-Next: How to Run Qwen’s 80B Sparse MoE Coder — Qwen3-Coder-Next uses sparse MoE and hybrid attention to enable efficient, long-context coding agent capabilities.
• What is trending in Hugging Face on Microsoft Foundry? Feb, 2, 2026 — Trending Hugging Face models are now production-ready in Microsoft Foundry, offering advanced AI capabilities.
• Best Small Language Models on Hugging Face Right Now! — Small language models now rival larger models in performance, enabling local, cost-effective deployment.
• Now in Foundry: Qwen3.5 Medium Model Series — Qwen3.5 models offer diverse VLM capabilities, from low-latency dense to cost-efficient MoE architectures.
• Qwen 3.7 Max & Plus First Test | Coding, Game Dev, OCR, Image Understanding Testing — Qwen 3.7 Max and Plus models show improved performance and reduced 'overthinking' compared to previous iterations.
• Qwen3.6-27B: Flagship-Level Coding in a 27B Dense Model — Qwen3.6-27B offers flagship coding performance in a significantly smaller, dense model.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.