# API Key Documentation for Coding Tools

Source: https://inference.raindragon.ai/coding-tools (Markdown: https://inference.raindragon.ai/coding-tools.md; index of all docs: https://inference.raindragon.ai/llms.txt)

Configure your API key and base URL for popular command-line and IDE tools

## Speed

Measured through our public API, the way your tool calls it: streaming, a fresh prompt each time (no cache hits), up to 512 output tokens.

### Gemma 4 26B (`google/gemma-4-26B-A4B-it`)

|  | 1 request at a time | 4 requests at once |
| --- | --- | --- |
| Output speed per request | ~100 tokens/s | ~85 tokens/s |
| Time to first token (~1K-token prompt) | ~2–3 s | ~2 s |

- A ~6K-token prompt still starts answering in about 3 s, then streams at ~80 tokens/s.
- For a typical coding-agent turn, a 500-token reply streams in about 5 seconds.

**Gemma 4 31B** (`google/gemma-4-31B-it`) is our larger dense model. It starts just as fast (~2 s) but writes at about 15 tokens/s. Use it when you want its quality, and 26B when speed matters.

Measured October 2026. Real speed varies with load and prompt length.

## OpenCode

OpenCode is an open-source coding agent for the terminal. Add RainDragon as a custom provider in `opencode.json` (in your project, or `~/.config/opencode/opencode.json` for every project).

```
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "raindragon": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "RainDragon",
      "options": {
        "baseURL": "https://inference.raindragon.ai/v1",
        "apiKey": "{env:RAINDRAGON_API_KEY}"
      },
      "models": {
        "google/gemma-4-26B-A4B-it": {
          "name": "Gemma 4 26B A4B",
          "limit": { "context": 204800, "output": 32768 }
        },
        "google/gemma-4-31B-it": {
          "name": "Gemma 4 31B",
          "limit": { "context": 65536, "output": 16384 }
        }
      }
    }
  }
}
```

Set your key, start OpenCode, and pick a RainDragon model with `/models`:

```
export RAINDRAGON_API_KEY="your-api-key"
opencode
```

Instead of the environment variable you can run `/connect`, choose Other, enter the provider ID `raindragon`, and paste your key.

## Hermes Agent

Hermes Agent by Nous Research works with any OpenAI-compatible endpoint. The quickest way is the setup wizard:

```
hermes model
# choose "Custom endpoint", then enter:
#   API base URL: https://inference.raindragon.ai/v1
#   API key:      your-api-key
#   Model name:   google/gemma-4-26B-A4B-it
```

Or set it directly in `~/.hermes/config.yaml`:

```
model:
  default: google/gemma-4-26B-A4B-it
  provider: custom
  base_url: https://inference.raindragon.ai/v1
  key_env: RAINDRAGON_API_KEY
```

and export the key in the shell that runs Hermes:

```
export RAINDRAGON_API_KEY="your-api-key"
```

## Cline

Cline, a VS Code extension, allows you to configure API keys and custom base URLs directly within its settings.

1. Open the Cline extension settings in VS Code.
2. Select "OpenAI Compatible" from the API Provider dropdown.
3. Enter your API key into the OpenAI API Key field:

```
API Key: your-api-key
```

1. Check the "Use custom base URL" box and enter the base URL:

```
Base URL: https://inference.raindragon.ai/v1
```

## OpenHands

OpenHands is an open-source AI agent that uses a web-based user interface for configuration.

1. Launch the OpenHands application and navigate to the web UI, typically at http://localhost:3000.
2. Go to the Settings page and find the LLM tab.
3. Toggle Advanced options to reveal additional settings.
4. In the Base URL field, specify:

```
Base URL: https://inference.raindragon.ai/v1
```

1. Set your API key:

```
API Key: your-api-key
```

## Aider Coder

Aider can connect to any LLM with an OpenAI-compatible API endpoint.

### Environment Variables

You can set the base URL and API key using environment variables:

```
export OPENAI_API_BASE=https://inference.raindragon.ai/v1
export OPENAI_API_KEY="your-api-key"
```

### YAML Config File

You can also add these settings to your .aider.conf.yml file:

```
openai-api-base: https://inference.raindragon.ai/v1
openai-api-key: "your-api-key"
```

## Codex CLI

Codex CLI allows you to define custom providers in a configuration file.

1. Create or edit the configuration file located at ~/.codex/config.toml.
2. Define a new model provider block:

```
[model_providers.raindragon]
name = "RainDragon AI"
base_url = "https://inference.raindragon.ai/v1"
env_key = "RAINDRAGON_API_KEY"
```

1. Then, set the environment variable RAINDRAGON_API_KEY with your actual key:

```
export RAINDRAGON_API_KEY="your-api-key"
```

## Claude Code

While Claude Code is optimized for Anthropic's models, it can be configured to use a custom base URL by editing a local configuration file.

1. Create or edit the settings.json file in your .claude directory, typically located at ~/.claude/settings.json.
2. Add the following env block to the file, replacing "your-api-key" with your actual key:

```
{
  "env": {
    "ANTHROPIC_BASE_URL": "https://inference.raindragon.ai/",
    "ANTHROPIC_API_KEY": "your-api-key"
  }
}
```

## Easy LLM CLI

Easy LLM CLI, a fork of the Gemini CLI, explicitly supports custom LLM providers through environment variables.

Set the following environment variables to configure your custom provider:

```
export USE_CUSTOM_LLM=true
export CUSTOM_LLM_PROVIDER="openai"
export CUSTOM_LLM_API_KEY="your-api-key"
export CUSTOM_LLM_ENDPOINT="https://inference.raindragon.ai/v1"
export CUSTOM_LLM_MODEL_NAME="google/gemma-4-26B-A4B-it"
```

## Qwen Code

Qwen Code CLI, another fork of the Gemini CLI, also uses environment variables to configure an OpenAI-compatible API with a custom base URL.

Set the environment variables with your custom provider's details:

```
export OPENAI_API_KEY="your-api-key"
export OPENAI_BASE_URL="https://inference.raindragon.ai/v1"
export OPENAI_MODEL="google/gemma-4-26B-A4B-it"
```

Generate API keys in your dashboard under API Settings.