From the source
Microsoft's MAI is releasing a family of seven models it says match or beat rival systems at a fraction of the price, aimed at enterprises whose token spend is climbing as they run agents around the clock.
- MAI Cyber 2 Flash is built for security work and, Microsoft says, matches GPT-6 Astra's quality at less than 2% of the cost per task on the CyberGymE2E benchmark, with output tokens priced 86% lower than MAI Cyber 1.1 Flash.
- MAI Code 2 reaches 98% of GPT-6.1 Sol's coding quality at about 40% of the price and 96% of Opus 5.5's performance at 5% of the price, though Microsoft says the models are not yet tuned for fundamental science research or very long horizon agentic work.
- MAI Code 2 Flash targets high-volume tasks at 65% of GPT-6 Luna's price and 20% of Haiku 5.5's, and is available in
GitHub Copilot and in a limited access preview in Microsoft Foundry.
- MAI Image 2.7 takes the top spot on Artificial Analysis image editing, ahead of GPT Image 2.6 Sunburst (max) at one fifth of the price, while MAI Realtime 1, the company's first live speech-to-speech model, starts at $0.01 per audio minute.
- Dragon Copilot Transcribe, built on MAI Transcribe 2, is live in Dragon Copilot and covers 10 million encounters a month, and Decision 1.1 is 50x faster and 79x cheaper than GPT-Sol 6.1.






