At Ignite 2026 in San Francisco, Microsoft AI launched seven in-house MAI models in one sweep — the company's biggest push yet to reduce its reliance on outside model providers. The lineup covers coding, security, image editing, speech and transcription.
The seven models are: MAI Code 2 and MAI Code 2 Flash for coding, now generally available in GitHub Copilot; MAI Cyber 2 Flash for security; MAI Image 2.7 for image editing, which Microsoft claims ranks #1 on Artificial Analysis; MAI Realtime 1 for speech-to-speech; Dragon Copilot Transcribe; and Decision 1.1.
Microsoft also made bold efficiency claims for its flagship coder: the company says MAI Code 2 delivers 98% of GPT-6.1 Sol's coding quality at 40% of the cost, and 96% of Claude Opus 5.5's coding quality at 5% of the cost. These are Microsoft's own benchmark claims, not independently verified, so treat the numbers as the company's pitch rather than settled fact.
Alongside the models, Copilot itself is being reorganised around four modes — Chat, Cowork, Code and Autopilot — as Microsoft moves from a single assistant to a set of specialised agents. The message from Ignite is clear: Microsoft wants its own models powering its own products, at a fraction of the cost of the frontier alternatives.