Documentation

Cline (VS Code extension)

Cline has first-class support for OpenAI-compatible APIs.

1. Open VS Code → Cline sidebar → Settings (gear icon).
2. Set API Provider to OpenAI Compatible.
3. Set Base URL to https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1.
4. Set API Key to your Unified AI key.
5. Set Model to any ID from the models table — e.g. deepseek-v4-flash.
6. Click Save.

# Equivalent cline settings.json fragment
{
  "cline.apiProvider": "openai",
  "cline.openAiBaseUrl": "https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1",
  "cline.openAiApiKey": "YOUR_KADEGATE_KEY",
  "cline.openAiModelId": "deepseek-v4-flash"
}

OpenCode

OpenCode reads provider config from ~/.config/opencode/config.json.

{
  "providers": {
    "Unified AI": {
      "type": "openai",
      "baseUrl": "https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1",
      "apiKey": "YOUR_KADEGATE_KEY"
    }
  },
  "model": "Unified AI/deepseek-v4-flash"
}

Then launch OpenCode — it will pick up the provider automatically. Use opencode --model "Unified AI/deepseek-v4-flash" to override at runtime.

Continue (VS Code / JetBrains)

Add a custom model block to ~/.continue/config.json:

{
  "models": [
    {
      "title": "Unified AI",
      "provider": "openai",
      "model": "deepseek-v4-flash",
      "apiBase": "https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1",
      "apiKey": "YOUR_KADEGATE_KEY"
    }
  ]
}

Reload VS Code / the IDE. The model will appear in the Continue model picker. You can add multiple entries — one per model you want to switch between.

OpenAI CLI / shell

Point the official openai Python CLI at Unified AI with two environment variables:

export OPENAI_API_BASE=https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1
export OPENAI_API_KEY=YOUR_KADEGATE_KEY

# Chat
openai api chat_completions.create \
  -m deepseek-v4-flash \
  -g user "Hello, world"

Or with the newer openai v1 SDK:

from openai import OpenAI

client = OpenAI(
    base_url="https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1",
    api_key="YOUR_KADEGATE_KEY",
)

r = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello"}],
)
print(r.choices[0].message.content)

Aider

Aider supports OpenAI-compatible endpoints via --openai-api-base:

aider \
  --openai-api-base https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1 \
  --openai-api-key YOUR_KADEGATE_KEY \
  --model deepseek-v4-flash

Or set it permanently in ~/.aider.conf.yml:

openai-api-base: https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1
openai-api-key: YOUR_KADEGATE_KEY
model: deepseek-v4-flash

Aider will use streaming automatically. All models in the list above are compatible.

OpenClaw

OpenClaw uses a YAML config file. Add Unified AI as a backend:

# ~/.openclaw/config.yaml
backends:
  - name: Unified AI
    type: openai_compatible
    base_url: https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1
    api_key: YOUR_KADEGATE_KEY
    default_model: deepseek-v4-flash
active_backend: Unified AI

Start OpenClaw — it will route all requests through Unified AI. Switch models with /model <id> inside a session.

Hermes Agent

Hermes Agent reads provider settings from its config or environment:

export HERMES_OPENAI_BASE_URL=https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1
export HERMES_OPENAI_API_KEY=YOUR_KADEGATE_KEY
export HERMES_MODEL=deepseek-v4-flash
hermes run

Or in hermes.config.json:

{
  "provider": "openai",
  "openai": {
    "baseUrl": "https://transcribe.h0yx2rtus9gfkf.flashpanel.link/api/v1",
    "apiKey": "YOUR_KADEGATE_KEY",
    "model": "deepseek-v4-flash"
  }
}

Hermes supports tool calling and streaming — both work transparently through Unified AI.