-
Notifications
You must be signed in to change notification settings - Fork 123
Parallel completions
igardev edited this page Jan 5, 2026
·
3 revisions
Llama-vscode generates parallel code completions (default 3) if a version of llama.cpp after December, 6, 2025 (commit c42712b) is used. The next completion is shown by pressing Ctrl+], previous completion is shown by pressing Ctrl+[.
The setting max_parallel_completions determines how many completions are generated.
- Run the completion model and start coding
- When a code completion is shown, press Ctrl+] to show the next completion, Ctrl+[ to show the previous completion
- Alternatively - you could hover over the shown completion and when the toolbar is shown click the arrows to show the other completions.
Settings:
- max_parallel_completions: The max number of parallel completions to generate. Default is 3.