Deploy and Host GPT Academic on Railway

GPT Academic is a Gradio web workbench for large language models aimed at research work: paper translation and summarisation (PDF and LaTeX), polishing, code explanation and a set of function plugins.

About Hosting GPT Academic

This template runs the project's ghcr.io/binary-husky/gpt_academic_nolocal:master image (no local models). When you deploy, enter API_KEY (an OpenAI key; several keys can be comma-separated). The UI is protected by a login: the user is admin and the password is the generated ADMIN_PASSWORD variable on the service. The default model is LLM_MODEL (gpt-4o-mini); change it or other options with environment variables named like the settings in config.py, e.g. API_URL_REDIRECT for an OpenAI-compatible endpoint.

Common Use Cases

Translate and summarise English papers and LaTeX projects
Polish and proofread academic writing
A shared, password-protected LLM workbench for a small group

Dependencies for GPT Academic Hosting

One stateless service
An OpenAI (or compatible) API key

Deployment Dependencies

Why Deploy GPT Academic on Railway?

Railway gives you a public HTTPS URL and automatic restarts with no server to maintain. Enter your key at deploy time, then log in with the generated password.