Skip to content
Free since 2015. Built for the engineers who keep things running.

cat /articles/microsofts-new-coding-model-promises-near-opus-performance-at-5-of-the-cost.md

AI & MLOps

Microsoft’s new coding model promises near-Opus performance at 5% of the cost

Microsoft unveiled two coding models, MAI Code 2 and a lighter Flash variant, claiming near-frontier results on coding benchmarks at a fraction of the price. The piece also weighs those claims against fresh rival releases.

The New Stack17 Nov 2026 · 3 min read · summary

article --summary

The summary

Microsoft used its Ignite conference to launch MAI Code 2 and MAI Code 2 Flash, two coding models it says come close to leading systems from OpenAI and Anthropic while costing far less to run. The company's headline comparisons put Code 2 at roughly 98 percent of one OpenAI model's performance for about 40 percent of the cost, and around 96 percent of an Anthropic model's performance for roughly 5 percent of the cost. Flash is pitched as matching Anthropic's and OpenAI's smaller models on several coding benchmarks at a lower price. The article points out that these claims arrived alongside an OpenAI release that narrows the gap, so the comparisons may look less impressive in a few days. It also highlights Microsoft's strategy of tuning narrow, efficient models for everyday developer work rather than the hardest research or long-running agent tasks, which Microsoft admits these models are not yet built for. Readers interested in the economics of AI coding tools or Microsoft's shift toward specialized models will find the most value here. The models are available now inside GitHub Copilot, with Flash in limited preview on Microsoft Foundry, and other platforms such as OpenRouter and Vercel are expected to add support soon.

Editorial note: This is a summary of an article first published by The New Stack. The full piece is on their site.

Reading now

Topic
AI & MLOps
Source
The New Stack
Published
17 Nov 2026
Format
Curated summary

Have a take on this?

Write it up and submit it. It goes on the list once an editor approves it.

Submit an article

articles --latest

Latest articles

New from the blogs we follow and the community.

All articles

newsletter --weekly

DevOps Weekly, minus the noise.

The best reads, new jobs and upcoming events, in one email a week.