Dify Third-Party API Quick Start: Configure an OpenAI-Compatible Model Provider

Dify supports third-party model services through OpenAI-compatible providers. Because NixAPI exposes a unified OpenAI-compatible API, you can add it to Dify and use it in chat apps, agent apps, and workflow nodes.

This guide follows the shortest path: create a key, add the provider, enter the Base URL and model name, then test a Dify app.

---

Prerequisites

You need:

• Access to a Dify console
• A NixAPI API Key
• A Dify app or workflow for testing

If you do not have a key, create one in the NixAPI console. Model list and prices are available on the NixAPI pricing page.

---

Step 1: Confirm the Base URL

For Dify’s OpenAI-compatible provider, use:

https://nixapi.com/v1

Dify will use this Base URL to call APIs such as chat/completions and embeddings.

Do not enter https://api.openai.com/v1, or requests will go to the official endpoint. Do not omit /v1.

---

Step 2: Install the OpenAI-Compatible Provider

In the Dify dashboard, open:

Settings → Model Providers

Find OpenAI-API-compatible or a similarly named provider plugin. If Dify asks you to install or update it, choose the latest version.

After installation, click Add Model.

---

Step 3: Fill in the Model Configuration

Use:

Field	Example	Notes
API Key	nix-your-api-key	Copy from the NixAPI console
Base URL	https://nixapi.com/v1	OpenAI-compatible endpoint
Model name	claude-sonnet-4-20250514	The model Dify should use

Suggested models:

• claude-sonnet-4-20250514
• gpt-5.4
• gemini-3-pro-preview
• deepseek-v4
• minimax-m3

The model name must match a supported NixAPI ID. Copy it from the NixAPI supported models page.

Dify will usually run a connectivity check after saving. If it fails, check the key, Base URL, and model name first.

---

Step 4: Test in an App

Open an existing chat app or create a test app. Select the NixAPI model you just added.

Enter:

Rewrite this sentence in a more formal tone: Today we should finish API integration first.

If Dify returns a normal answer, the provider setup works.

In logs, you may see a request like:

POST https://nixapi.com/v1/chat/completions
Authorization: Bearer nix-your-api-key

That confirms Dify is calling NixAPI.

---

Step 5: Use It in Workflow

In Dify Workflow, choose the configured NixAPI model in an LLM node.

Common patterns:

• User question → knowledge retrieval → NixAPI model summary
• Form input → NixAPI model structured JSON output
• HTTP request node → NixAPI model report generation

If your workflow calls models repeatedly or processes long text, check the NixAPI model pricing page. NixAPI currently has a limited-time top-up exchange rate: RMB 0.80 = USD 1.00.

---

Optional: Configure Embeddings

If your Dify app uses knowledge-base retrieval, you can add an embeddings model through the same OpenAI-compatible provider.

Use:

Field	Example
API Key	nix-your-api-key
Base URL	https://nixapi.com/v1
Model name	Use a supported embeddings model from NixAPI

Then select that embeddings model in the knowledge-base settings.

---

Troubleshooting

401 Unauthorized

The API Key is wrong or was not saved correctly. Copy it again from the NixAPI console.

Timeout

Check that the Base URL is:

https://nixapi.com/v1

If Dify runs on a server, make sure the server can access that URL.

JSON parsing error

This is often caused by prompt or response-format settings. First test a normal chat app, then enable JSON mode, structured output, or function calling step by step.

Model name error

Copy the exact model ID from the NixAPI supported models page.

---

Summary

Dify third-party API setup requires the OpenAI-API-compatible provider, https://nixapi.com/v1, a valid API Key, and an exact model ID. Once configured, chat apps, agent apps, and workflows can reuse the same model provider.