Fast mode
Fast mode swaps the main model for a lighter, quicker one with a single keypress, and swaps back just as easily. Use it when speed matters more than depth: small edits, quick questions, renaming things, running the same chore across many files.
Turn it on and off
| How | Effect |
|---|---|
/fast | Toggles fast mode now. The status bar confirms it, for example Fast mode enabled (model: Syntic Code Fast). |
Alt+O | Same as /fast. You can rebind it as chat:fastMode in keyboard shortcuts. |
"fastMode": true in a settings file | Every session starts in fast mode. You can also flip Fast mode in /config. |
While fast mode is on, the footer shows [FAST], and a status line script receives "fast_mode": true in its input.
/fast takes effect from the next request. To have it apply straight away, even in the middle of a reply, set SYNTIC_IMMEDIATE_COMMAND=1.
Which model it uses
When you turn fast mode on, Syntic Code remembers the model you were using and switches to the fast one:
- Your
smallModel(from settings,--small-modelorSYNTIC_SMALL_MODEL), if the Syntic platform offers it to the CLI and it is not the model you are already on. - Otherwise, Syntic Code Fast.
When you turn fast mode off, the model you had before comes back. Picking a model in /model while fast mode is on ends fast mode and uses the model you picked.
Fast mode only changes the model. Your reasoning effort, permission mode, rules, memory and tools stay the same.
When to use it
| Good fit | Better without fast mode |
|---|---|
| Small, well-described edits | Debugging something you do not understand yet |
| Questions about code you can point at | Designing a change across many files |
| Repetitive changes across files | Security-sensitive changes |
| Formatting, renames, boilerplate | Long tasks where a wrong turn is costly |
A rule of thumb: if you could describe the exact change in one sentence, fast mode is usually enough.
Fast mode or lower effort?
Both make replies quicker. They work differently:
| Fast mode | Lower effort | |
|---|---|---|
| What changes | The model | How much the same model reasons before answering |
| Turn on | /fast | /effort low |
| Good for | Simple tasks overall | Simple steps inside a task that still needs the main model |
You can combine them. Effort is remembered per model, so the level you last used with the fast model comes back when you switch to it. See Models and reasoning.
Cost
The fast model uses fewer resources per request, so routine work in fast mode uses less of your plan. Use /cost or /usage to compare: costs are listed per model. See Usage and billing.
Availability
The model list comes from the Syntic platform. If your plan does not include the fast model, /model shows it with the plan it needs, and requests are refused with that information. An organization can also keep it off the list with availableModels or deniedModels in managed settings.
Subagents choose their own model. Fast mode changes the main conversation only. To run subagents on the fast model, set SYNTIC_SUBAGENT_MODEL or name the model in the agent’s definition.
Related pages
- Models and reasoning: picking models, effort and thinking.
- The advisor: a second opinion from another model.
- Plans and models: which models your plan includes.