Tune GitHub Copilot Inline Suggestions with Configurable Eagerness
GitHub Copilot in VS Code now offers a configurable eagerness setting that controls how often inline suggestions appear and how large they are. Low, medium, and high modes all run on one model, selected with a control tag. The setting is available to paid Copilot subscribers, so you can match the assistance level to your task.
Key points
- Eagerness has three modes: low, medium, and high; medium is the default.
- Low mode offers fewer, smaller and more conservative edits.
- High mode suggests more proactively and can propose larger edits when they fit the task.
- Update to the latest VS Code and confirm Next Edit Suggestions are enabled, which is the default.
- The feature is for paid GitHub Copilot subscribers and is chosen from the Copilot settings menu.
Some developers found Copilot's inline suggestions too frequent, while others wanted it to reach further ahead. The VS Code team responded with a three-level eagerness setting. Low eagerness keeps suggestions smaller and more conservative, high eagerness offers more proactive and sometimes larger edits, and medium keeps today's default behavior.
The team measured eagerness with edit distance, the number of characters a suggestion changes, and trained one model that switches behavior based on a control tag rather than shipping three separate models. Training combined supervised fine-tuning with on-policy distillation and reinforcement learning, which lets the same approach carry over to future base models. Reward hacking, such as padding high-eagerness edits with unrelated changes or minifying low-eagerness code, was caught and fixed. In online tests, high eagerness raised accepted characters, and low eagerness reduced dismissals.
Real-world use
- Writing repetitive patterns quickly with high eagerness to get several steps ahead.
- Working through delicate business logic with low eagerness so each suggestion stays scoped to the current statement.
- Scaffolding a new class where high eagerness can anticipate validation and error handling.
Why it matters
You can now decide how much Copilot intervenes while you type, reducing distracting suggestions or speeding up familiar patterns. Your choice is a simple setting rather than a model swap.