GitHub Copilot: HydraFusion Gets a Better Model Pool and New Max Option
GitHub CopilotGitHub updated HydraFusion, the research preview that orchestrates multiple models into a single Copilot response, with a refreshed model pool, per-step reasoning effort, and improved Cascade and Critique workflows. GitHub also introduced HydraFusion Max, a tier tuned for frontier capability on demanding tasks. Both options are available in VS Code, the GitHub Copilot app, and GitHub Copilot CLI for Pro, Pro+, Business, and Enterprise plans.
Key Takeaways
- HydraFusion Max adds a higher-capability tier aimed at demanding tasks, alongside the balanced standard HydraFusion option.
- The model pool now includes Claude Opus 5.5, GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna, mixing providers within a single response.
- Per-step reasoning effort runs medium effort for drafts and high effort for review, replacing each model's default setting.
- Cascade can now add a judging step that requests a repair only when needed, reducing wasted escalation.
- Critique now inspects the workspace with tools and uses parallel critic models for more grounded reviews.
- Both options are still a research preview, available in VS Code, the Copilot app, and Copilot CLI on Pro, Pro+, Business, and Enterprise plans.
Sources & Mentions
4 external resources covering this update
What HydraFusion Does
HydraFusion is a GitHub Copilot research preview that orchestrates multiple models to produce a single response, so developers no longer need to choose a model for every task. GitHub announced the original Project HydraFusion in September 2026, and the October 14 update refines how it picks and combines models while adding a second, more capable option.
Updated Model Pool and Workflows
The HydraFusion model pool now draws on Claude Opus 5.5, GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna. Which models run for a given task depends on the orchestration workflow HydraFusion selects.
Reasoning Effort per Step
HydraFusion now sets reasoning effort for each step of a workflow instead of relying on each model's default. Initial drafts run at medium effort, while review steps run at high effort. This lets the system spend more thinking time where it improves quality and less where a draft only needs to be good enough to review.
Cascade and Critique Improvements
The Cascade workflow can now include a judging step that requests a repair only when needed, which avoids unnecessary escalation to a stronger model. The Critique workflow now leans more heavily on tools to inspect the workspace directly, and it uses parallel critic models to review the work.
New HydraFusion Max Option
GitHub added HydraFusion Max, built on the same architecture and workflow techniques as HydraFusion but tuned for frontier capability on demanding tasks. It applies higher reasoning effort where depth matters while still taking token efficiency into account.
The two options are positioned differently:
- HydraFusion: balanced, with moderate token savings, for tasks that need a mix of speed, cost, and performance.
- HydraFusion Max: for tasks that need the deepest reasoning and the most thorough analysis.
Availability
Both HydraFusion and HydraFusion Max remain in research preview. They are available in VS Code, the GitHub Copilot app, and GitHub Copilot CLI. They are offered on Copilot Pro, Pro+, Business, and Enterprise plans with experimental or preview features enabled, subject to organization model policies. The announcement includes no pricing details. GitHub points users to the HydraFusion documentation and a community feedback discussion for sharing results.