Apr 24, 2026 1.4 KB markdown of v3

modal

run python functions on cloud GPUs with modal on sublimated. write a _run.py, click run, get results back in your folder.

how it works

  • create a folder and add a _run.py file — sublimated detects it as a programmable folder

  • _run.py reads files from the folder, processes them, writes output files back

  • click the run button on any function in the toolbar — runs on Modal with GPU/CPU as configured

  • output files appear in your folder automatically via git push

what's supported

  • any python code (pandas, PIL, matplotlib, torch, etc)

  • GPU acceleration (T4, A10G, A100, H100) via modal decorators

  • custom pip dependencies via requirements.txt

  • multiple functions per file (each gets its own run button)

  • automatic file I/O — your code reads/writes files in the current directory

  • output files pushed back to your folder after execution

  • real-time status via activity indicators

demos

working examples with source files:

  • combine-text — combines text files into one output (plain python, no modal)

  • csv-analysis — reads CSV files, produces summary stats and pivot tables

  • image-resize — creates thumbnails and contact sheets from images

  • json-to-chart — generates PNG bar charts from JSON data

guide

step-by-step tutorial: guide


faq · all docs