Skip to main content
Editor settings control how code cells look and behave.

​
Code completion

Choose how code completion works: LM completion considers notebook context — nearby cells and their outputs — to make relevant suggestions. Choose an autocomplete profile:
  • Fast uses Qwen3 Coder Next and prioritizes low latency.
  • Smart uses GPT-5.6 Luna Fast and prioritizes completion quality.
Fast is the default. Accounts without access to the Fast model are served Smart instead. Both profiles require a Quadrillion API key and are served through the Quadrillion proxy, so they are unavailable on BYOK deployments. Autocomplete never uses your own provider keys, even when other OpenAI requests are configured to use them.

​
Display options

​
Line numbers

Show or hide line numbers in code cells:
  • Enabled: Numbers appear in the left gutter
  • Disabled: Clean view without numbers
Toggle with Shift+L in command mode.

​
Line wrapping

Control how long lines display:
  • Enabled: Lines wrap to fit the cell width
  • Disabled: Lines extend with horizontal scroll

​
LaTeX rendering

Enable math rendering in markdown cells:
  • Uses KaTeX for fast rendering
  • Supports inline ($...$) and block ($$...$$) math

​
Search settings

​
Max search results

Set the upper limit for search results:
  • Higher values show more results
  • Very high values may slow the interface
  • Default balances completeness and performance
This affects both in-document find and workspace search.

​
Data display

​
CSV rows per page

When viewing CSV files, control pagination:

​
Execution time format

How cell execution times display:
  • Compact: “1.2s”
  • Human-readable: “1 second 234 milliseconds”
Execution times appear below completed cells.