Quickstart

1. Get a key

Sign up at runaii.cloud (you get $5 in free credits), then create a key at Console → API Keys. Keys start with sk-runaii- and are shown once — store them somewhere safe.

2. Point your SDK at us

The API is OpenAI-compatible — if you already use the OpenAI SDK, this is a two-line change:

from openai import OpenAI

client = OpenAI(
    api_key="sk-runaii-...",                    # your runaii key
    base_url="https://api.runaii.cloud/v1",     # ← the only change
)

resp = client.chat.completions.create(
    model="deepseek-v4-flash",         # or any id from /api/v1/models
    messages=[{"role": "user", "content": "Hello, fleet."}])

print(resp.choices[0].message.content)
print(resp.usage.cost)                          # exactly what you were charged

deepseek-v4-flash is one of the models the live catalog currently reports as ready. glm-5.3 reports is_ready: false at the time of writing, and calls against a false lane will not serve — check is_ready in GET /api/v1/models before wiring a model into production.

3. Or plain curl

curl https://api.runaii.cloud/v1/chat/completions \
  -H "Authorization: Bearer $RUNAII_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Hello, fleet."}]
  }'

4. Anthropic SDK? Same models, same billing

The Anthropic /v1/messages dialect is served natively — point the Anthropic SDK at our /api base and keep using x-api-key exactly as you do with Claude:

import anthropic

client = anthropic.Anthropic(
    api_key="sk-runaii-...",                     # your runaii key
    base_url="https://api.runaii.cloud/api",     # ← the only change
)

msg = client.messages.create(
    model="deepseek-v4-flash",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello, fleet."}],
)
print(msg.content[0].text)

Both dialects are the same product underneath: tool use (tools / tool_choice / tool_result round-trips), image blocks, top_k, and thinking → reasoning are all supported on the Anthropic path, and the same per-key rate limits, daily caps, and per-token billing apply either way. Coding agents (Claude Code, Cline, Cursor, …) have per-tool setup snippets on the connect page.

5. Responses API? Also native (Codex CLI, Agents SDK)

The OpenAI Responses API (/v1/responses) is served natively too — point Codex CLI, the OpenAI Agents SDK, or any Responses-based client straight at it:

resp = client.responses.create(
    model="deepseek-v4-flash",
    input="Hello, fleet.",
)
print(resp.output_text)

curl https://api.runaii.cloud/v1/responses \
  -H "Authorization: Bearer $RUNAII_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "input": "Hello, fleet."
  }'

You get the full Responses surface: input (string, message items, image parts), instructions, max_output_tokens, function tools with function_call / function_call_output round-trips, text.format (json_object / json_schema), reasoning, and complete response.* streaming events. Two things are not supported, by design: previous_response_id (the endpoint is stateless — send the full conversation in input, which is what the SDK does when store: false) and OpenAI-hosted tools (web_search, file_search, computer_use, code_interpreter) — use type: "function" tools instead, which work everywhere. The same key, the same rate limits, the same per-token billing as the other dialects. Non-streaming responses report usage.cost exactly like chat completions.

All three dialects — OpenAI chat completions, Anthropic messages, and the Responses API — are one product underneath: authenticate with the same sk-runaii- key and pick whichever shape your app already speaks.

What you get back

{
  "choices": [{ "message": { "content": "Hello, fleet." } }],
  "usage": {
    "prompt_tokens": 14,
    "completion_tokens": 9,
    "total_tokens": 23,
    "cost": 0.000006
  }
}

usage.cost is exactly what your balance was charged for this request — metered per token, verified to the micro. Check your running balance any time with GET /api/v1/key.