Microsoft introduces MAI Code 2 and Code 2 Flash
Original title: MAI Code 2 and Code 2 Flash
Microsoft says Code 2 reaches 98% of GPT 6.1 Sol on four coding benchmarks at 40% of the cost.
Microsoft announced MAI Code 2 and MAI Code 2 Flash, coding models aimed at everyday enterprise work rather than long-horizon science or agentic research. Code 2 is positioned for production changes, debugging, multi-file work, repository reasoning, and UI, while Flash targets explanations, small edits, routine fixes, and fast iteration. Microsoft says that, averaged on DeepSWE 1.1, Terminal Bench 2.1, SWE-Atlas, and FrontierCode v1.1, Code 2 reaches 98% of GPT 6.1 Sol at 40% of the cost and 96% of Claude Opus 5.5 at 5% of the cost. It also says blind frontend reviewers preferred Code 2 over GPT-6 Sol 67% of the time, and that Flash matches GPT-6 Luna and Claude Haiku 5.5 at 65% and 20% of the cost.
Unconfirmed: Cost and benchmark figures are Microsoft’s own comparisons, not independent results.
- Microsoft says the models are not yet tuned for harder, long-horizon problems.
- Benchmarks cited are those Microsoft sees in GitHub Copilot usage.
- Flash is compared on three day-to-day, high-volume coding benchmarks.
- Frontend preference is from Microsoft’s blind app evaluations.
- Vendors
- Microsoft
- Products
- GitHub Copilot
From the original article
1 min read at microsoft.ai
: t wo new coding models built for the everyday enterprise coding tasks , delivering leading performance at a fraction of the cost of competing models. MAI Code 2 is at the frontier of price-performance for common enterprise coding tasks and excels in taking code to production, deep debugging, multi-file implementation, repo wide reasoning, and UI.
By lemonish97 · microsoft.ai
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.