Plug your agent into runaii.

Point Claude Code at runaii (env vars — set before launching):

$ export ANTHROPIC_BASE_URL=https://api.runaii.cloud/api
export ANTHROPIC_AUTH_TOKEN=sk-runaii-…   # your API key

Claude Code works against this endpoint: set the two env vars above and launch. The Anthropic-compatible route translates tool_use/tool_result and tool_choice, so the agentic loop runs. Use an sk-runaii API key (the console session path is text-only).

Model routing: /model glm-5.3 (tool-calling capable). glm models are reasoning models — max_tokens must be ≥ 512 (their thinking shares the budget; Claude Code's defaults are already high enough).

Settings → Models → OpenAI API key → override base URL:

$ # Cursor settings JSON (Models → OpenAI API Key → override)
{
  "apiKey": "sk-runaii-…",
  "baseUrl": "https://api.runaii.cloud/v1"
}

Enable the runaii models you want (they appear after Verify):

Add model IDs exactly as glm-5.3, qwen3.8-flash — no prefix.

Cline settings → API Provider → OpenAI Compatible:

$ Base URL:  https://api.runaii.cloud/v1
API Key:   sk-runaii-…
Model ID:  glm-5.3

~/.codex/config.toml — point the OpenAI provider at runaii:

$ [model_providers.runaii]
name = "runaii"
base_url = "https://api.runaii.cloud/v1"
env_key = "RUNAII_API_KEY"
 
# then: profile.default_provider = "runaii"

One env var pair — aider reads OpenAI-compatible endpoints natively:

$ export OPENAI_API_BASE=https://api.runaii.cloud/v1
export OPENAI_API_KEY=sk-runaii-…
aider --model openai/glm-5.3

~/.zcode/cli/config.json — add a custom provider, then pick it as your model:

$ {
  "provider": {
    "runaii": {
      "name": "runaii.cloud",
      "kind": "openai",
      "options": {
        "baseURL": "https://api.runaii.cloud/v1",
        "apiKey": "sk-runaii-…"
      },
      "enabled": true,
      "source": "custom",
      "models": {
        "glm-5.3": { "name": "glm-5.3 — agentic + coding" }
      }
    }
  },
  "model": "runaii/glm-5.3"
}

Model IDs are bare (glm-5.3, qwen3.8-flash) — same list as GET /api/v1/models.

opencode.json (project root or ~/.config/opencode/) — custom provider:

$ {
  "provider": {
    "runaii": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "runaii.cloud",
      "options": {
        "baseURL": "https://api.runaii.cloud/v1",
        "apiKey": "sk-runaii-…"
      },
      "models": {
        "glm-5.3": {}
      }
    }
  }
}

Then: opencode --model runaii/glm-5.3 (provider-id/model-id — the runaii/ prefix is opencode's provider key, not part of the API model id).

Set both base-URL names — the OpenAI SDK reads OPENAI_BASE_URL; aider and LiteLLM read OPENAI_API_BASE:

$ export OPENAI_BASE_URL=https://api.runaii.cloud/v1
export OPENAI_API_BASE=https://api.runaii.cloud/v1
export OPENAI_API_KEY=sk-runaii-…

Anthropic-dialect tools: ANTHROPIC_BASE_URL=https://api.runaii.cloud/api + ANTHROPIC_AUTH_TOKEN — supports text, images, and tool-use with an API key.