Microsoft unveils MAI Code 2 AI coding models, claims near-Opus performance cheaply
At Ignite 2026, Microsoft introduced two in-house AI coding models, MAI Code 2 and Code 2 Flash, built for everyday coding tasks. Microsoft says MAI Code 2 delivers 96% of Claude Opus 5.5's performance at about 5% of its API price, with Code 2 Flash positioned against GPT-6 Luna and Claude Haiku 5.5. The models are being offered as optional alternatives within GitHub Copilot, Microsoft 365, and Fabric.
GoKawiil's interpretation of the reporting above, not reported fact.
These benchmarks suggest Microsoft is building its own competitive alternatives to the OpenAI and Anthropic models it has long relied on, which could reduce its dependence on outside AI providers and lower its own operating costs. The comparison is based on API usage pricing rather than consumer subscription costs, so everyday Copilot users likely won't see a direct price cut. Microsoft's push to benchmark itself against OpenAI, its largest AI partner, could also signal a shifting balance in that relationship.
- Microsoft launched MAI Code 2 and Code 2 Flash, new in-house coding AI models, at Ignite 2026.
- Microsoft claims 96% of Claude Opus 5.5's performance at roughly 5% of its API cost.
- The models are being integrated as optional choices in GitHub Copilot, Microsoft 365, and Fabric, potentially cutting Microsoft's reliance on OpenAI and Anthropic.
Source: zdnet.com — David Gewirtz, 2026-11-17
Published there as: “Microsoft says its new AI coding model rivals Claude Opus 5.5 at 5% of the cost”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.