GitHub Copilot's HydraFusion Orchestrates Multiple Models Per Task
WirelessLife · x · 2026-10-21
GitHub Copilot has launched HydraFusion in research preview. It's not a model but a runtime model orchestrator: selected from the model picker, it frames each task as an optimization problem and uses capability signals (reasoning, code generation, debugging, tool use) to pick the execution pattern, reasoning effort, and models.
Three execution patterns are currently supported:
- Single: one model completes the task
- Cascade: an efficient model drafts, a judge accepts or escalates to a stronger model
- Critique: one model drafts, another critiques, and the original revises once
Unlike auto model selection, HydraFusion also decides the execution pattern and per-step reasoning effort. It's research preview with no SLA and not intended for production use.
More from coding & agent
- Mastercard's Agent Pay for Machines goes live on x402, letting AI agents pay for services — kleffew94 · 2026-10-21
- Grok Bot gets X Money: chat agent can now check balances, send money and pay merchants — Scobleizer · 2026-10-21
- How do you evaluate GenAI output quality before shipping? Devs weigh in — ssnuopi · 2026-10-21
- RunInfra launches $10/mo coding plan: fastest open models across 19 agents incl. Claude Code — ycombinator · 2026-10-21
- Matt Shumer launches an 'AI employee' you assign work to over email — mattshumer_ · 2026-10-21
- uv-ty integration: how to enable PEP 723 script type checking — charliermarsh · 2026-10-21