> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Deep Research

> Learn how to configure Deep Research with ScitiX Model Inference using proxy mode and an OpenAI-compatible provider.

[Deep Research](https://github.com/u14app/deep-research) is a self-hosted web app that runs iterative, multi-stage research workflows. It supports **OpenAI Compatible** providers, so it can run on the ScitiX Model Inference API.

Default model here is `glm-5.2`; see the [Models](https://console.scitix.ai/model-inference/models) page for the full list and pricing.

Deploy it per the [official guide](https://github.com/u14app/deep-research) (e.g. the `xiangfa/deep-research` Docker image on port 3000).

## Configure

Deep Research has two **API Modes**: *Local* (the browser calls the model API directly) and *Proxy* (the app's server calls it). Use **Proxy** for ScitiX. In Local mode the browser calls `https://api.scitix.ai/...` directly (cross-origin); the ScitiX gateway does not return CORS headers for browser requests, so the call is blocked with `error [TypeError]: Failed to fetch`. This is a browser CORS limitation, not a key/URL mistake. Proxy mode makes the request from the app's server (same-origin to the browser), which avoids it.

Create an API key on the ScitiX [API Keys](https://console.scitix.ai/model-inference/api_keys) page and set the server-side environment (Proxy mode reads the key and endpoint from here):

```bash theme={null}
OPENAI_COMPATIBLE_API_KEY=<Your API Key>
OPENAI_COMPATIBLE_API_BASE_URL=https://api.scitix.ai/model-api
ACCESS_PASSWORD=<a password you choose>
```

| Field                            | Value                                                                                                   |
| -------------------------------- | ------------------------------------------------------------------------------------------------------- |
| `OPENAI_COMPATIBLE_API_KEY`      | Your ScitiX API key                                                                                     |
| `OPENAI_COMPATIBLE_API_BASE_URL` | `https://api.scitix.ai/model-api` without `/v1`; the proxy route already appends `/v1/chat/completions` |
| `ACCESS_PASSWORD`                | A password you choose for the Deep Research UI                                                          |

Then in the UI open **Settings → Model** and set:

| Field                       | Value                                        |
| --------------------------- | -------------------------------------------- |
| API Mode                    | **Proxy**                                    |
| AI Provider                 | **OpenAI Compatible**                        |
| Access Password             | The `ACCESS_PASSWORD` you set                |
| Thinking Model / Task Model | `glm-5.2`. Set both to a tool-capable model. |

<img src="https://mintcdn.com/scitix-adfc65c2/AblxxMM8pLmVYmdt/model-inference/media/deep-research-proxy-config.png?fit=max&auto=format&n=AblxxMM8pLmVYmdt&q=85&s=41c357643a0e48a5f954a6b1f70b14f5" alt="Deep Research — Proxy mode with the OpenAI Compatible provider and glm-5.2" width="1032" height="970" data-path="model-inference/media/deep-research-proxy-config.png" />

## Verify

Enter a research topic and start — Deep Research runs its stages (topic → questions → plan → collection → report) using glm-5.2 through ScitiX:

<img src="https://mintcdn.com/scitix-adfc65c2/AblxxMM8pLmVYmdt/model-inference/media/deep-research-result.png?fit=max&auto=format&n=AblxxMM8pLmVYmdt&q=85&s=5a339bfb1933a4b4f9cab7d98afc6a6a" alt="Deep Research generating questions and a report plan via ScitiX glm-5.2" width="2098" height="3530" data-path="model-inference/media/deep-research-result.png" />

Verified on Deep Research v0.11.0.