Connect your tool

The three values#

# Any OpenAI-compatible tool: Cline, OpenCode, Roo…
base_url = https://api.inferiq.run/v1
api_key  = iq_…
model    = inferiq/balanced

Cline#

Cline talks to any OpenAI-compatible endpoint, so it works with Inferiq as-is.

1. Open Cline’s settings and set API Provider to “OpenAI Compatible”.
2. Paste the base URL and your Inferiq key.
3. Set Model ID to an alias such as inferiq/balanced, or an exact model id from Models.

Base URL   https://api.inferiq.run/v1
API Key    iq_…
Model ID   inferiq/balanced

Use inferiq/fast for Plan mode and inferiq/balanced for Act mode if you want to save on planning turns.

OpenCode#

Add Inferiq as a custom OpenAI-compatible provider in opencode.json.

1. Put your key in an environment variable: export INFERIQ_API_KEY=iq_…
2. Add the provider block below to opencode.json in your project (or ~/.config/opencode/).
3. Pick the model with /models inside OpenCode.

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "inferiq": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Inferiq",
      "options": {
        "baseURL": "https://api.inferiq.run/v1",
        "apiKey": "{env:INFERIQ_API_KEY}"
      },
      "models": {
        "inferiq/fast": { "name": "Inferiq fast" },
        "inferiq/balanced": { "name": "Inferiq balanced" },
        "inferiq/best": { "name": "Inferiq best" }
      }
    }
  },
  "model": "inferiq/inferiq/balanced"
}

OpenCode names models provider/model, so the alias inferiq/balanced from the provider “inferiq” becomes inferiq/inferiq/balanced.

Cursor#

Cursor can send chat to a custom OpenAI base URL.

1. Open Cursor Settings → Models.
2. Under OpenAI API Key, paste your Inferiq key and turn on “Override OpenAI Base URL”.
3. Set the base URL below, then add inferiq/balanced as a custom model and select it in chat.

OpenAI API Key              iq_…
Override OpenAI Base URL    https://api.inferiq.run/v1
Custom model                inferiq/balanced

Some Cursor features, such as Tab completion and parts of Agent mode, always use Cursor’s own models and ignore a custom base URL. Chat with your selected model goes through Inferiq.

Claude Code#

Claude Code speaks the Anthropic Messages format, which Inferiq serves at /v1/messages. Requests are answered by Qwen3.8-27B through the same aliases, limits and receipts.

1. Set ANTHROPIC_BASE_URL to Inferiq (without /v1; Claude Code adds the path).
2. Set ANTHROPIC_AUTH_TOKEN to your Inferiq key.
3. Pick the alias with ANTHROPIC_MODEL, and the one for quick background tasks with ANTHROPIC_SMALL_FAST_MODEL.

export ANTHROPIC_BASE_URL="https://api.inferiq.run"
export ANTHROPIC_AUTH_TOKEN="iq_…"
export ANTHROPIC_MODEL="inferiq/balanced"
export ANTHROPIC_SMALL_FAST_MODEL="inferiq/fast"

If Claude Code asks for a Claude model name, Inferiq maps it to the alias with the same depth: haiku → inferiq/fast, sonnet → inferiq/balanced, opus → inferiq/best. A key limited to certain models must allow the aliases Claude Code uses.

curl#

Check your key and the endpoint from a terminal.

1. Export your key.
2. Send one request. The response includes the model that answered and a receipt id.

export INFERIQ_API_KEY="iq_…"

curl https://api.inferiq.run/v1/chat/completions \
  -H "Authorization: Bearer $INFERIQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "inferiq/balanced", "messages": [{"role": "user", "content": "Say hi"}]}'

MCP server#

Inferiq is also a remote MCP server at https://api.inferiq.run/v1/mcp, so Claude Code, Cursor and other MCP clients can price and run fixes and check your balance without leaving the editor. There’s nothing to install. Use an API key with the fix scope; every fix still gets a quote first, is paid from your balance and gets a receipt.

claude mcp add --transport http inferiq https://api.inferiq.run/v1/mcp \
  --header "Authorization: Bearer iq_…"

{
  "mcpServers": {
    "inferiq": {
      "url": "https://api.inferiq.run/v1/mcp",
      "headers": { "Authorization": "Bearer iq_…" }
    }
  }
}

Tools

• inferiq_quote: price a fix before it runs.
• inferiq_fix: run an approved quote and return the cause, a patch, steps to verify and any safety warnings. Nothing is applied for you.
• inferiq_balance: your available balance and what’s held.

If it doesn’t work#