Copilot Fusion: less model-picking, a closer look at cost
In brief
According to Silicon.fr, GitHub has merged two Copilot features into a single Fusion mode that both picks a model and organizes the steps of a task. For each request, the system chooses how to proceed and at which of four effort levels. High and Max are limited to certain plans, with a 30% discount on model prices until the end of November.
Source : Silicon.fr - « GitHub Copilot fusionne finalement routeur et orchestrateur »
What it changes for an SME
If you already pay for Copilot, Fusion can save your developers time spent choosing a model. The bill still needs watching: Michelin, as reported by Silicon.fr, did not see the orchestrator as a substitute for Auto mode but as a way to raise quality, at a clearly higher price, and it often picked more powerful models than needed unless cheaper options were requested. High and Max effort levels are available only on Pro+, Max, Business and Enterprise. Standard remains available on every Copilot plan.
What you can do
- Ask your Copilot admin whether Fusion is on by default and which effort level applies to new conversations.
- Track model spend for a few weeks, especially if you are on Business or Enterprise.
- Check with developers whether they get enough progress information while tasks run, a gap already noted in early tests.
Let's talk about the time you could save.
In 20 minutes, we find where AI could save you time or money. Free, no commitment.