Google and Microsoft Bring New Models to Coding Tools

Avatar photo

ByRyan Mitchell

November 19, 2026

Gemini 4 Neon and Microsoft’s MAI Code 2 Flash are reaching developers through familiar coding tools, with different access, pricing and rollout details.

Google and Microsoft are bringing new models into tools developers already use, making it easier to test alternatives in existing workflows. Google launched Gemini 4 Neon on November 18. Microsoft’s MAI Code 2 Flash is available in GitHub Copilot and in limited-access preview through Microsoft Foundry. For software businesses, the practical question is whether these models perform well on their own code and at a cost that makes sense—not which company makes the largest benchmark claim.

Gemini 4 Neon has a one-million-token context window and four thinking levels. Google lists API prices of $3 per million input tokens, 15 cents per million cached input tokens and $15 per million output tokens. The model is rolling out through Google AI Studio, Gemini for business, Antigravity and the Gemini app. Those published rates give developers a basis for estimating costs, although actual bills will depend on how much context a task uses and how much output it produces.

Google says Neon ranks first overall on the Artificial Analysis Coding Agent Index, with a score of 58.6. It reports scores of 65.2% on Terminal-Bench 4.1 and 46.2% on FrontierCode 2.0 at the default high setting. The results make Neon a candidate for hands-on testing, not a guarantee of better results on every codebase. Developers should compare it on representative work, such as debugging, multi-file changes and tasks involving a large repository.

Neon became generally available in GitHub Copilot on November 18 for Pro+, Max, Business and Enterprise users. GitHub is gradually rolling it out across Visual Studio Code, Visual Studio, Copilot CLI, GitHub’s coding agent and website, mobile, JetBrains IDEs, Xcode and Eclipse. That range makes it relevant to teams with developers using different environments. Copilot administrators can control access through model policies.

The billing arrangement is a key caveat: GitHub says Neon usage is charged at Google’s list pricing, rather than covered by a separate fixed Copilot entitlement. Being able to select a model inside Copilot does not mean its usage is included at no additional cost. Businesses should confirm the applicable rates and monitor consumption before expanding access. The discounted cached-input rate may help workflows that reuse context, but savings depend on how those workflows are configured.

Microsoft’s MAI Code 2 Flash is available in GitHub Copilot and in limited-access preview in Microsoft Foundry. Microsoft plans to distribute it through OpenRouter, Vercel, Hermes, OpenCode and Kilo. Those additional channels could give developers more ways to evaluate the model beyond Microsoft’s own products, but the announcement does not specify when they will arrive or what their pricing and access terms will be.

The products are at different stages. Google has published Neon’s API rates and is rolling it out across consumer, business and developer products. Microsoft’s Flash is available in Copilot and preview in Foundry, while the other distribution channels remain planned. Developers should check the terms of each service rather than assume access or pricing carries over from one platform to another.

The larger change is that model choice is moving closer to the editor, terminal and cloud console, reducing the effort required to try alternatives. That can improve competition, but it also makes provider-specific billing and controls more important. Small businesses can start with a limited pilot, compare quality and speed on real tasks, and track usage before making a model broadly available. Neon is worth testing for teams interested in its large context window and published API pricing; MAI Code 2 Flash is a practical candidate for teams already using Copilot or Foundry. Neither launch proves which model will be the better choice across all workloads.

Leave a Reply

Your email address will not be published. Required fields are marked *