Using Nexhina in AI Coding Tools

This document explains how to integrate Nexhina into mainstream AI coding tools such as Cursor, Windsurf, and Continue, so that these tools can use your own models and quotas.

1. Using Nexhina in Cursor

Cursor is one of the most popular AI coding tools today and supports custom compatible APIs. Nexhina is fully compatible with the standard interface format and is ready to use after configuration.

1.1 Adding Nexhina Models

Open Cursor and go to Settings → Models

In the Models Names input field at the bottom, enter the model name you want to use and click Add model

Recommended models:

Use Case	Model Name	Description
Daily coding	`gpt-4o`	Top-tier multimodal, strong general coding
Complex reasoning	`o3`	Enhanced reasoning, good for debugging, math, and logic
General conversation	`gpt-4o-mini`	Fast and cost-effective, great for everyday chat
Long context	`gpt-4.1`	1M context window, handles large codebases
Professional writing	`claude-sonnet-4-20250514`	Claude Sonnet 4, strong at long-form generation and analysis

After adding, toggle on the switch for the corresponding model in the list

1.2 Configuring API Key and Base URL

In the same Settings → Models page, find the API Key configuration area:

Config Item	Value
API Key	`sk-your-Nexhina-key` (your Nexhina token)
Base URL	`https://token.nexhina.ai/v1`

Click Verify after filling in. A success message indicates the configuration is complete.

1.3 Using Models in Cursor

After configuration:

Open the Cursor Chat panel (shortcut Cmd+L / Ctrl+L)

In the model selection dropdown, select the model you just added

Chat normally — all requests will go through Nexhina

1.4 Configuring @Docs Document Context

Cursor's @Docs feature can inject external documents as context into the conversation, allowing the model to answer questions based on your API documentation.

Configuration steps:

Open Cursor Settings → Features → Docs

Click Add new doc

Fill in the configuration:

Config Item	Value
Name	`Nexhina Docs`
URL	`https://token.cshuacai.cn` (your documentation site URL)
Start URL (optional)	Documentation home page URL

Click Save

1.5 Referencing Documents with @Docs

In Cursor Chat:

Type @Docs and select Nexhina Docs

Then enter your question, for example:

@Nexhina Docs How do I use Function Calling?

@Nexhina Docs Help me write a Python function to call the Embedding API

Cursor will automatically fetch documentation content as context, and the model will give accurate answers based on the documentation.

2. Using Nexhina in Windsurf

Windsurf (formerly Codeium) also supports standard compatible APIs.

Configuration Steps

Open Windsurf Settings → AI Provider

Select OpenAI Compatible or Custom Provider

Fill in the configuration:

Config Item	Value
API Base URL	`https://token.nexhina.ai/v1`
API Key	`sk-your-Nexhina-key`
Model	`gpt-4o` or other available models

After saving, you can use it in Cascade and Chat

Windsurf Configuration File

You can also directly edit the configuration file ~/.windsurf/settings.json:

{
  "aiProvider": "openai-compatible",
  "openaiCompatible": {
    "baseUrl": "https://token.nexhina.ai/v1",
    "apiKey": "sk-your-Nexhina-key",
    "models": [
      { "id": "gpt-4o", "name": "GPT-4o" },
      { "id": "o3", "name": "o3" },
      { "id": "gpt-4o-mini", "name": "GPT-4o mini" },
      { "id": "claude-sonnet-4-20250514", "name": "Claude Sonnet 4" }
    ]
  }
}

3. Using Nexhina in Continue

Continue is an open-source AI coding assistant that supports VS Code and JetBrains.

Configuration Steps

Edit the Continue configuration file ~/.continue/config.json (or ~/.continue/config.yaml):

JSON Format

{
  "models": [
    {
      "title": "Nexhina GPT-4o",
      "provider": "openai",
      "model": "gpt-4o",
      "apiBase": "https://token.nexhina.ai/v1",
      "apiKey": "sk-your-Nexhina-key"
    },
    {
      "title": "Nexhina o3",
      "provider": "openai",
      "model": "o3",
      "apiBase": "https://token.nexhina.ai/v1",
      "apiKey": "sk-your-Nexhina-key"
    },
    {
      "title": "Nexhina GPT-4o mini",
      "provider": "openai",
      "model": "gpt-4o-mini",
      "apiBase": "https://token.nexhina.ai/v1",
      "apiKey": "sk-your-Nexhina-key"
    },
    {
      "title": "Nexhina Claude Sonnet 4",
      "provider": "openai",
      "model": "claude-sonnet-4-20250514",
      "apiBase": "https://token.nexhina.ai/v1",
      "apiKey": "sk-your-Nexhina-key"
    }
  ],
  "tabAutocompleteModel": {
    "title": "Nexhina Autocomplete",
    "provider": "openai",
    "model": "gpt-4o-mini",
    "apiBase": "https://token.nexhina.ai/v1",
    "apiKey": "sk-your-Nexhina-key"
  },
  "embeddingsProvider": {
    "provider": "openai",
    "model": "text-embedding-3-large",
    "apiBase": "https://token.nexhina.ai/v1",
    "apiKey": "sk-your-Nexhina-key"
  }
}

YAML Format

Configuration Notes

Field	Description
`models`	List of conversation models, appears in Continue's model selection dropdown
`tabAutocompleteModel`	Code completion model, recommended to use a fast model (gpt-4o-mini)
`embeddingsProvider`	Embedding model for codebase indexing

Restart VS Code / JetBrains after configuration to select Nexhina models in the Continue panel.

4. Using Nexhina in VS Code Copilot

GitHub Copilot supports third-party APIs through the custom model feature in Copilot Chat.

Configuration Steps

Install the GitHub Copilot and GitHub Copilot Chat extensions

Open VS Code Settings → search for github.copilot.chat

Configure custom endpoints (requires VS Code 1.90+ and Copilot custom model support)

Configuring via Environment Variables

Set environment variables in the terminal before starting VS Code:

Note: Copilot's support for custom models is continuously evolving, and the specific configuration method may change between versions. If custom models are not supported, we recommend using Cursor or Continue as alternatives.

5. Using Nexhina in JetBrains AI

The AI Assistant in JetBrains IDEs (IntelliJ IDEA / PyCharm / WebStorm, etc.) supports custom endpoints.

Configuration Steps

Open Settings → Tools → AI Assistant → Providers

Select OpenAI Compatible or Custom Provider

Fill in:

Config Item	Value
Server URL	`https://token.nexhina.ai/v1`
API Key	`sk-your-Nexhina-key`
Model	`gpt-4o` or other

Click Test Connection to verify

Save and you can use it in AI Assistant

6. Using in Chat Tools

If you only want to use Nexhina's models in a chat interface (rather than coding tools), you can do so in the following ways:

Using Third-Party Clients

Any client that supports custom APIs can be used:

Client	Platform	Configuration
ChatBox	Desktop	Settings → API Base URL + Key
NextChat	Web	Settings → Endpoint + Key
LobeChat	Web/Desktop	Settings → Model Service → Proxy URL + Key
Open WebUI	Web	Settings → API URL + Key
Cherry Studio	Desktop	Settings → API URL + Key

Common configuration:

Config Item	Value
API Base URL	`https://token.nexhina.ai/v1`
API Key	`sk-your-Nexhina-key`

7. FAQ

Q: Verify fails in Cursor. What should I do?

Check the following:

Check Item	Correct Value
Base URL	`https://token.nexhina.ai/v1` (ends with `/v1`)
API Key	Starts with `sk-`, no extra spaces
Model name	Must be an id returned by `GET /v1/models`, watch out for case sensitivity
Network connectivity	`curl https://token.nexhina.ai/v1/models` should return normally

Q: Models don't appear in Cursor?

Make sure the model toggle is on

Quit and reopen Cursor

Check that the model name is spelled correctly (all lowercase, e.g. gpt-4o not GPT-4o)

Q: Code completion is slow?

Code completion is latency-sensitive, so we recommend:

Use a fast model: gpt-4o-mini

Avoid using reasoning models (o3) for completion

Continue users can configure a fast model separately in tabAutocompleteModel

Q: Model returns `model_not_found` error in chat?

The model is not enabled under your Nexhina account. Contact the administrator to enable it, or switch to an available model.

Yes, but be aware:

All tools share the Key's quota, so watch the usage

Concurrent requests share the Key's rate limit

We recommend using different Keys for different tools to make management and monitoring easier

Q: Can I configure other services and Nexhina at the same time?

Cursor: Does not support configuring two endpoints simultaneously; the later configuration will override. If you need to use both, we recommend splitting them through Continue or another tool.

Continue: Supported. Add configurations with different apiBase in the models array.

Q: @Docs configuration doesn't reference any content?

Make sure the documentation URL is publicly accessible

Try opening the configured URL in a browser to confirm the page works

If it's an internal documentation site, Cursor may not be able to crawl it

Q: Continue's Embedding index fails?

Make sure embeddingsProvider is configured correctly:

{
  "provider": "openai",
  "model": "text-embedding-3-large",
  "apiBase": "https://token.nexhina.ai/v1",
  "apiKey": "sk-your-Nexhina-key"
}

If the error persists, check that the Key has permission for the Embedding API.

Configuration Quick Reference

There are only two core configurations for all tools:

Config Item	Value
Base URL	`https://token.nexhina.ai/v1`
API Key	`sk-your-Nexhina-key`

Configuration entries for each tool:

Tool	Configuration Entry	Description
Cursor	Settings → Models → API Key	Fill in Base URL + Key
Windsurf	Settings → AI Provider	Select Compatible
Continue	`~/.continue/config.json`	Edit config file
JetBrains	Settings → Tools → AI Assistant	Select Custom Provider
ChatBox	Settings → API	Fill in URL + Key
LobeChat	Settings → Model Service	Fill in Proxy URL + Key
NextChat	Settings → Endpoint	Fill in URL + Key
Open WebUI	Settings → API	Fill in API URL + Key

Table of Contents#

1. Using Nexhina in Cursor#

1.1 Adding Nexhina Models#

1.2 Configuring API Key and Base URL#

1.3 Using Models in Cursor#

1.4 Configuring @Docs Document Context#

1.5 Referencing Documents with @Docs#

2. Using Nexhina in Windsurf#

Configuration Steps#

Windsurf Configuration File#

3. Using Nexhina in Continue#

Configuration Steps#

JSON Format#

YAML Format#

Configuration Notes#

4. Using Nexhina in VS Code Copilot#

Configuration Steps#

Configuring via Environment Variables#

5. Using Nexhina in JetBrains AI#

Configuration Steps#

6. Using in Chat Tools#

Using Third-Party Clients#

7. FAQ#

Q: Verify fails in Cursor. What should I do?#

Q: Models don't appear in Cursor?#

Q: Code completion is slow?#

Q: Model returns model_not_found error in chat?#

Q: Can multiple tools share the same Key?#

Q: Can I configure other services and Nexhina at the same time?#

Q: @Docs configuration doesn't reference any content?#

Q: Continue's Embedding index fails?#

Configuration Quick Reference#

Table of Contents

1. Using Nexhina in Cursor

1.1 Adding Nexhina Models

1.2 Configuring API Key and Base URL

1.3 Using Models in Cursor

1.4 Configuring @Docs Document Context

1.5 Referencing Documents with @Docs

2. Using Nexhina in Windsurf

Configuration Steps

Windsurf Configuration File

3. Using Nexhina in Continue

Configuration Steps

JSON Format

YAML Format

Configuration Notes

4. Using Nexhina in VS Code Copilot

Configuration Steps

Configuring via Environment Variables

5. Using Nexhina in JetBrains AI

Configuration Steps

6. Using in Chat Tools

Using Third-Party Clients

7. FAQ

Q: Verify fails in Cursor. What should I do?

Q: Models don't appear in Cursor?

Q: Code completion is slow?

Q: Model returns `model_not_found` error in chat?

Q: Can multiple tools share the same Key?

Q: Can I configure other services and Nexhina at the same time?

Q: @Docs configuration doesn't reference any content?

Q: Continue's Embedding index fails?

Configuration Quick Reference