HydraFusion research preview reaches VS Code and the Copilot app

GitHub’s changelog for 30 September 2026 says the HydraFusion research preview is now in Visual Studio Code and the GitHub Copilot app, after an earlier preview that was limited to Copilot CLI. HydraFusion shows up in the model picker, but the post is explicit that it is not a model. It treats the choice of workflow as an optimization problem and uses signals for reasoning, code generation, debugging, and tool use to pick a pattern.

The three workflows

The changelog names three:

Single: one selected model solves the task.

Cascade: a cheaper model drafts an answer, and a quality gate either accepts it or escalates to a stronger model.

Critique: one model drafts, an independent read-only critic from a different model family reviews the draft, and the drafting model revises once. GitHub says this follows the same review pattern as Rubber Duck.

The post says this release was the top request from early users, and that the new surfaces also show more of what HydraFusion is doing, send progress updates more often, and make a long task look active rather than stuck. It distinguishes HydraFusion from Auto: Auto picks a model for each request, while HydraFusion picks a workflow and can coordinate more than one model inside a single turn.

Who can turn it on

HydraFusion is available to Copilot Pro, Pro+, Business, and Enterprise. Business and Enterprise administrators have to enable preview features before it appears. In VS Code the post requires version 1.140 or later, or VS Code Insiders. You select HydraFusion in the Copilot Chat model picker. If it is missing, the setting to enable is chat.copilot.hydraFusion.enabled. In the Copilot app, update to the latest version, search settings for HydraFusion, turn it on, and then pick it in the model picker.

GitHub says the feature remains a research preview and is subject to change. The changelog points to HydraFusion documentation and to a separate write-up, “Project HydraFusion: Frontier quality via multi-model orchestration,” for research and benchmark results. The benchmark numbers live in that write-up, not in the changelog.

Practical takeaway

Turn it on only if you want a preview that may change, and only on Pro or above, with an admin switch for company accounts. Expect one of three workflows rather than a new base model, and use the progress UI to see which path ran. If you need a number for quality against a single model, read the research post GitHub links; this changelog does not contain that table.