Microsoft’s new Licensed Weights lets businesses customize AI models with their own data
ZDNET’s key takeaways
- Microsoft will let customers customize its new models’ weights.
- The approach should appeal to businesses on privacy, ability, and cost.
- Western open-weight AI is trending upward.
During the keynote at its Ignite conference on Tuesday, Microsoft CEO Satya Nadella announced several new models from Microsoft AI (MAI), its lab within the company. But new model releases are almost a dime a dozen these days, and accelerating fast — what really stood out was how Microsoft will let its customers access those models.
Also: Microsoft says its new AI coding model rivals Claude Opus 5.5
MAI said it would allow its customers to customize its frontier models with their own data and run them securely in Azure through an initiative called Licensed Weights. It’s essentially a selective approach to open-weight AI: more so than proprietary models from OpenAI and Anthropic, but not fully downloadable on Hugging Face like Mistral’s Le Chonk. Interested companies will need to go through their Microsoft account reps to get started.
“You control who can access it. Your data and adaptations stay entirely yours. And we’ll work with you through Frontier Company to deliver the custom model and embed it in your own products and agents,” Mustafa Suleyman, CEO of MAI, said during the keynote.
The approach should stand out to businesses looking for affordable AI that lets them retain control over their IP. Microsoft said it would only use customers’ information to train its models if they agreed.
Open-weight AI for enterprises
The dominant, if simplified, narrative in AI has been that proprietary, expensive models boast peak capabilities but aren’t customizable, while open, cheaper models are customizable but less capable. Microsoft’s AI models don’t necessarily compete with OpenAI and Anthropic yet, but the company has legacy sway with countless businesses. By offering a customizable AI option, it could strike a balance among major tech companies.
Western labs, including Thinking Machines and Mistral, are releasing more competitive open-weight models, but Anthropic and OpenAI still have the overwhelming majority of the enterprise AI market. Though Licensed Weights isn’t an open-weight family of models, it does diversify the open-versus-closed debate a bit and differentiate Microsoft’s offering from OpenAI’s and Anthropic’s. It’s also a smart pairing with Microsoft’s goal to balance price and performance, if businesses find it impressive enough.
Also: Microsoft unveiled Copilot Workspaces at Ignite 2026
“Averaged over four widely used benchmarks that target the everyday enterprise coding tasks we see in GitHub Copilot, MAI Code 2 gives you 98% of the performance of GPT 6.1 Sol at just 40% of the cost, and against Opus 5.5 it’s an incredible 5% of that cost,” Suleyman said of the new coding model the company announced at Ignite.
The company also released its new image editing model, MAI Image 2.7, on Tuesday, which now leads the category on Artificial Analysis while beating Nano Banana Pro and several GPT image models on price and quality. With Licensed Weights, Microsoft has positioned itself for organizations seeking a reliable blend of ROI tools, customization, security controls, and capability.
Radhika Rajkumar
Senior Editor
Radhika Rajkumar is a senior editor at ZDNET based in New York City. She covers AI, specializing in safety, privacy and security, policy, education, and synthetic media. She also leads ZDNET's newsletter strategy. Radhika holds a Masters in Creative Publishing and Critical Journalism from The New School. See full bio
What's Your Reaction?
Like
0
Dislike
0
Love
0
Funny
0
Wow
0
Sad
0
Angry
0
Comments (0)