Copy the example file and edit it with your configuration:
Variables marked Yes under Required must be set for DeepTutor to start. Optional variables enable additional features or override defaults.
LLM settings
The primary language model used for all AI operations — chat, problem solving, research, idea generation, and more.
| Variable | Required | Default | Description |
|---|
LLM_BINDING | Yes | openai | Provider binding. Options: openai, azure_openai, anthropic, deepseek, openrouter, groq, together, mistral, ollama, lm_studio, vllm, llama_cpp |
LLM_MODEL | Yes | — | Model name, e.g. gpt-4o, deepseek-chat, claude-3-5-sonnet |
LLM_API_KEY | Yes | — | API key for the LLM provider |
LLM_HOST | Yes | — | API endpoint URL, e.g. https://api.openai.com/v1 |
LLM_API_VERSION | No | — | API version string. Required for Azure OpenAI, e.g. 2024-02-15-preview |
Embedding settings
The embedding model powers the RAG (Retrieval-Augmented Generation) pipeline used by the knowledge base and search features.
| Variable | Required | Default | Description |
|---|
EMBEDDING_BINDING | Yes | openai | Provider binding. Options: openai, azure_openai, jina, cohere, huggingface, google, ollama, lm_studio |
EMBEDDING_MODEL | Yes | text-embedding-3-small | Embedding model name |
EMBEDDING_API_KEY | Yes | — | API key for the embedding provider |
EMBEDDING_HOST | Yes | — | API endpoint URL, e.g. https://api.openai.com/v1 |
EMBEDDING_DIMENSION | Yes | 3072 | Vector dimensions. Must match the model’s output dimensions |
EMBEDDING_API_VERSION | No | — | API version string. Required for Azure OpenAI |
The demo knowledge bases shipped with DeepTutor were built with text-embedding-3-large at dimensions = 3072. If you use a different model or dimension, create a new knowledge base — you cannot mix embeddings from different models in one index.
Port and access settings
| Variable | Required | Default | Description |
|---|
BACKEND_PORT | No | 8001 | Port for the FastAPI backend |
FRONTEND_PORT | No | 3782 | Port for the Next.js frontend |
NEXT_PUBLIC_API_BASE | No | — | Frontend API URL. Set this for LAN or remote access, e.g. http://192.168.1.100:8001 |
NEXT_PUBLIC_API_BASE_EXTERNAL | No | — | External API URL for cloud or Docker deployments. Takes priority over NEXT_PUBLIC_API_BASE |
DISABLE_SSL_VERIFY | No | false | Disable SSL certificate verification. Not recommended for production |
Choosing the right API base variable
The frontend needs to know the backend’s address. The startup scripts resolve it using the following priority order:
NEXT_PUBLIC_API_BASE_EXTERNAL — use for cloud or Docker deployments where the backend is accessed via a public URLNEXT_PUBLIC_API_BASE — use for LAN or remote access from another device on the same network- Default:
http://localhost:{BACKEND_PORT} — works only when the browser and backend are on the same machine
For a simple home or office setup where you access DeepTutor from another device (e.g. 192.168.1.66:3782), set:NEXT_PUBLIC_API_BASE=http://192.168.1.66:8001
For any cloud deployment (Docker on a remote server, VPS, etc.), you must set NEXT_PUBLIC_API_BASE_EXTERNAL to your server’s public URL. Without it, the browser will try to contact localhost:8001, which will fail.
Search provider settings
Enable web search for the Deep Research and Smart Solver modules.
| Variable | Required | Default | Description |
|---|
SEARCH_PROVIDER | No | perplexity | Search provider. Options: perplexity, tavily, serper, jina, exa |
SEARCH_API_KEY | No | — | API key for the selected search provider |
SEARCH_API_KEY value is used for whichever provider SEARCH_PROVIDER is set to.
TTS settings
Enable audio narration in the Interactive IdeaGen (Co-Writer) module.
| Variable | Required | Default | Description |
|---|
TTS_BINDING | No | openai | Provider binding. Options: openai, azure_openai |
TTS_MODEL | No | tts-1 | TTS model name |
TTS_API_KEY | No | — | API key for the TTS provider. Can be the same as LLM_API_KEY when using OpenAI |
TTS_URL | No | — | API endpoint URL, e.g. https://api.openai.com/v1 |
TTS_VOICE | No | alloy | Voice for audio output. Options: alloy, echo, fable, onyx, nova, shimmer |
TTS_BINDING_API_VERSION | No | — | API version string. Required for Azure OpenAI |
HuggingFace / MinerU settings
These variables are optional and relate to the MinerU PDF parser used during knowledge base creation.
| Variable | Required | Default | Description |
|---|
HF_ENDPOINT | No | — | HuggingFace mirror endpoint, e.g. https://your-hf-mirror.example.com |
HF_HOME | No | — | HuggingFace cache directory. Recommended: /app/data/hf with a Docker volume mount to reuse downloaded models |
HF_HUB_OFFLINE | No | — | Set to 1 to force offline mode. Requires models already present in HF_HOME |