GitHub is expanding Copilot Fusion across major developer tools, promising near-frontier coding results through multi-model orchestration while using fewer tokens and costing less than rival systems.
GitHub is widening the pitch for Copilot Fusion, a new orchestration layer that decides how different models share a coding task. The feature is available in Visual Studio Code, the Copilot app, Copilot CLI and JetBrains IDEs, and it gives administrators control over which models can be used, as well as the option to make Fusion the default experience. According to the material provided by 4sysops, Fusion can run at Standard, High or Max effort levels, allowing it to assign work to one model, escalate an initial attempt to a stronger one, or route a draft through a second model for review and revision.
The most notable claim is that Fusion is intended to narrow the gap with top-tier coding systems without matching their full cost. Artificial Analysis measured Fusion Max at an average of $5.59 per task, compared with $11.42 for Claude Code running Opus 5.5 Max. On the same benchmark, Fusion scored 60 on the Coding Agent Index, two points behind the rival system, while using roughly half as many tokens and completing tasks about 40% faster, according to the figures cited in the report.
That pricing model may matter as much as the benchmark result. There is no separate orchestration charge, but each model call is billed at the model’s standard rate, which means the total cost depends on how often Fusion escalates work between models. GitHub is also discounting Fusion model costs by 30% until 30 November 2026, a time-limited incentive that could make the feature more attractive to teams comparing it with other AI coding tools priced on a flat per-seat basis.
The wider Copilot ecosystem still matters here. Recent reviews of the product note that its broad IDE coverage remains one of its biggest strengths, with support across Visual Studio Code, JetBrains tools, Visual Studio, Neovim, Xcode, Eclipse, Zed and GitHub’s own interfaces. That reach, combined with agent mode and enterprise controls, helps explain why GitHub is positioning Fusion as an upgrade to an already familiar workflow rather than as a separate product category. In practical terms, the company is trying to sell a more efficient route to near-frontier coding performance without asking users to leave the tooling they already use.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
