Microsoft targets OpenAI and Anthropic with cheaper custom AI models
Microsoft is positioning its own AI models as a more cost-effective and customizable alternative to its partners' tools.

The short version
- Microsoft claims Copilot costs $30 for everyday users compared to $81 for ChatGPT and $205 for Claude.
- New MAI models offer 96% of Anthropic's coding performance at only 5% of the cost on specific benchmarks.
- Companies can now use MAI Licensed Weights to customize models with private data in their own Azure environments.
- Microsoft introduced new tools to help IT leaders track and cap AI spending to avoid exceeding budgets.
Why it matters
This signals a shift where Microsoft is moving from being just a distributor of other companies' AI to a primary provider of its own. For businesses, it means more options to balance high-end performance with strict budget constraints.
A shift toward competition
Microsoft is taking a more aggressive stance against its own partners, OpenAI and Anthropic. While the company remains a major investor and distributor for both firms, it recently positioned its own tools as superior alternatives for business customers. During the Ignite keynote, executives highlighted that Microsoft's offerings provide more power for less money.
Judson Althoff, CEO of Microsoft’s commercial business, stated that users can do anything in Claude or ChatGPT but do it better with Copilot. The company is trying to move the conversation away from just what AI can do toward how much it costs to actually run it at scale. This move comes as Anthropic begins to expand into Microsoft's core business areas with its own office-style tools.
The strategy involves leveraging the existing dominance of the Microsoft 365 suite. By integrating AI directly into the full versions of Word, Excel, and PowerPoint, Microsoft aims to provide a seamless experience. They want to convince IT professionals that using their native tools is more efficient than using the simplified viewers offered by competitors.
The cost of AI at scale
A major part of Microsoft's pitch focuses on the high costs of using third-party AI models for enterprise work. Microsoft commissioned research from GigaOm to compare the monthly costs for business users. Their data suggests that Copilot is significantly cheaper than the competition for both everyday and advanced users.
For an everyday user, Microsoft estimates the cost of Copilot at $30 per month. In contrast, they estimate ChatGPT costs $81 and Claude costs $205. These figures are based on enterprise pricing, which includes both a per-seat fee and additional charges based on how much the AI is actually used.
The price gap widens for advanced users who utilize agent tools. Microsoft estimates these costs at $96 for Copilot, while ChatGPT sits at $149 and Claude reaches $299. Most of the Claude figure is attributed to estimated usage charges on top of a $20 enterprise seat fee. Microsoft is using these numbers to argue that their platform is the most economical choice for large-scale deployments.
New MAI models and licensing
To support this cost-saving narrative, Microsoft released a new batch of its own MAI models. These models are designed to compete directly with OpenAI and Anthropic while maintaining a lower price point. One specific example is MAI Code 2, which is now available for use in GitHub Copilot.
Microsoft claims that its MAI models can achieve high performance levels at a fraction of the cost of rival models. For instance, they say MAI Code 2 reaches 96% of the coding power found in Anthropic’s Claude Opus 5.5 while costing only 5% of the price on specific benchmarks. They also introduced MAI Cyber 2 Flash, a security model that performs as well as OpenAI’s GPT-6 Astra on one benchmark but costs less than 2% per task.
Beyond just providing models, Microsoft is introducing a new business model called MAI Licensed Weights. This allows companies to take a Microsoft AI model and customize it with their own private data. These customized models can then be run within a company's own Azure environment with help from Microsoft's consulting arm.
Managing the AI budget
Microsoft is also addressing the reality that many companies are struggling to keep their AI spending under control. The company cited a McKinsey survey where 93% of respondents reported that they have exceeded their AI budgets. To solve this, they are introducing new management tools for IT leaders.
A new feature called Agent 365 Business Analytics will show leaders exactly what is driving their AI spending and what they are getting for that investment. Additionally, a companion Cost Management tool will allow IT departments to set specific spending policies and receive alerts when they approach certain limits. These tools are meant to provide transparency in a landscape where usage fees can fluctuate.
The goal is to provide a controlled environment where businesses can experiment with AI without fear of runaway costs. By offering both the models and the tools to monitor them, Microsoft is positioning itself as a comprehensive partner for enterprise AI. This approach aims to help companies achieve what CEO Mustafa Suleyman calls long-term self-sufficiency in their AI operations.
The road ahead for Copilot
The competition between Microsoft and its partners is becoming increasingly complex as products become more intertwined. For example, Anthropic's Claude for Microsoft 365 allows users to use Claude inside Office apps for tasks like sorting emails. However, internal data shows that this product has fewer weekly users than the paid version of Copilot in Office.
Microsoft views this disparity as a sign that they can successfully keep customers from switching to rival platforms. By keeping the core experience inside the Office suite, they create a unified environment that is harder for users to leave. This strategy aims to solidify their position as the primary gateway for enterprise AI applications.
Moving forward, the focus will likely remain on cost transparency and customization. As companies look to integrate AI into their daily workflows, Microsoft's ability to offer both high-performance models and the tools to manage their costs will be a key differentiator. This marks a significant evolution in how the company balances its roles as both a partner and a competitor in the AI space.
What to watch
- How many companies choose MAI Licensed Weights over standard API models.
- Whether Anthropic's new office tools can gain enough market share to challenge Copilot.
Try it
Check your company's current AI usage costs to see if you are exceeding your budget.
Sources and links
- Reportinggeekwire.com: the article we sourced
First reported by geekwire.com. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.