HudHud Tech — Tech & AI news platform
Wednesday, 18 November 2026
Breaking
Inventors & Creators | Artificial Intelligence

Microsoft Launches Seven MAI Models at Lower Costs

18 November 2026 3 min read هدهد تك | Hudhud Tech

Microsoft has unveiled seven new models under its MAI family, targeting coding, cybersecurity, image editing, speech, transcription, and automated decision-making. This launch aims to compete with OpenAI and Anthropic by offering advanced models at lower costs.

Announced on November 17 at its Ignite conference, the new lineup includes MAI Cyber 2 Flash, MAI Code 2, MAI Code 2 Flash, MAI Image 2.7, MAI Realtime 1, Dragon Copilot Transcribe, and Decision 1.1. Microsoft is focusing on making these advanced models more affordable for businesses that run automated agents continuously.

The company noted that rising token costs are making price-performance increasingly important for enterprises looking to scale their systems. Microsoft is positioning these models for specific workloads rather than relying on a single system for all tasks.

MAI Cyber 2 Flash is designed for continuous cybersecurity monitoring and defensive tasks, including identifying and patching vulnerabilities in live production systems. Microsoft claims it matches the quality of GPT-6 Astra at less than 2% of the cost per task on the CyberGymE2E benchmark.

The model builds on earlier MAI Cyber systems, with its predecessor accounting for nearly 40% of scans in Microsoft’s security assessment platform, MDASH. Additionally, 90% of scans that began with MAI continued using it.

For software developers, MAI Code 2 targets everyday enterprise coding tasks, delivering 98% of GPT-6.1 Sol’s coding performance at about 40% of the price. Compared to Anthropic’s Claude Opus 5.5, it claims to provide 96% of the performance at just 5% of the cost.

A lighter version, MAI Code 2 Flash, is aimed at routine tasks such as code explanations and targeted edits. It is available in GitHub Copilot and in limited preview through Microsoft Foundry.

Microsoft also introduced MAI Image 2.7, which ranks first in image editing according to Artificial Analysis. The company claims it outperforms OpenAI’s GPT Image 2.6 Sunburst at one-fifth of the price, with its text-to-image generation performance ranking fourth in the same evaluation.

MAI Realtime 1 is Microsoft’s first speech-to-speech model, designed for natural conversations with low latency, starting at $0.01 per audio minute. Dragon Copilot Transcribe, based on MAI Transcribe 2, powers 10 million patient encounters each month and has reportedly halved multilingual transcription error rates.

Decision 1.1 provides model routing within applications, directing tasks to larger models only when necessary. Microsoft claims it is 50 times faster and 79 times cheaper than GPT-Sol 6.1, currently used by 4,000 customers.

Additionally, Microsoft is introducing MAI Licensed Weights, allowing businesses to customize models with proprietary data and run them in their own Azure environments. The company highlighted Maia 200, its second-generation AI accelerator, which is in production and delivers up to 1.7 times more tokens per watt than leading GPU infrastructure.

By combining specialized models, customizable weights, and in-house accelerators, Microsoft aims to provide businesses with more control over how they deploy and pay for automated systems. This announcement comes as major technology companies compete to lower the costs of running increasingly capable models.

 

Source: Interesting Engineering

What do you think of this story?

Share this story: