What is a flat-rate OpenAI-compatible API?

Last updated October 26, 2026 by the APIClaw team

A flat-rate OpenAI-compatible API accepts standard OpenAI SDK requests but bills a fixed monthly price instead of per token. APIClaw is one: $19 to $129 a month buys a daily request quota, and you switch by changing the base URL to https://apiclaw.biz/v1 and the API key.

Definition

An OpenAI-compatible API is a service that speaks the same HTTP format as the OpenAI API, mainly Chat Completions and the Responses API. Any client written for OpenAI, including the official Python and JavaScript SDKs, works with it once you change the base URL and the API key. Most tools that let you "override the OpenAI base URL" are using this idea.

A flat-rate one adds a billing promise. You pay a fixed monthly price for a fixed allowance, and the cost does not rise with the number of tokens in each request. APIClaw is a flat-rate OpenAI-compatible gateway: the allowance is a daily request quota, the price runs from $19 to $129 a month, and the quota resets at 00:00 UTC. For a general overview of the idea, see the flat-rate AI API overview.

Request-based versus token-based pricing

Token pricing is fair when requests are small and rare. It becomes unpredictable when a tool resends a long conversation on every call, which is how coding agents behave. Request pricing trades that unpredictability for a ceiling you know in advance.

At low volume, a few dozen modest calls a day, per-token billing is often cheaper than $19 a month. A flat plan pays off once your usage is heavy and regular. The OpenRouter alternative guide shows how this compares with a per-token router that adds a credit fee.

How APIClaw plans and weights work

Each plan has a number of requests a day. Models count as one request or more, called their weight. The weight depends on the model, not on how many tokens you send, and there is no cap on tokens per request beyond the model's own limit. Models have up to 1M-token context.

Claude Sonnet 5.5 counts as 2 requests, so the $19 Plus plan covers 250 Sonnet calls a day. DeepSeek V4 Flash counts as 1. Cycles are 30 days and you can cancel any time. Larger needs are handled as a custom plan through Telegram support. The model hub lists every model and weight.

Python example

Install the official SDK, then point it at APIClaw:

from openai import OpenAI

client = OpenAI(
    base_url="https://apiclaw.biz/v1",
    api_key="sk-your-apiclaw-key",
)

response = client.chat.completions.create(
    model="claude-sonnet-5-5",
    messages=[{"role": "user", "content": "Explain request-based pricing in one sentence."}],
)
print(response.choices[0].message.content)

JavaScript example

The same change works with the OpenAI package for Node.js and edge runtimes:

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://apiclaw.biz/v1",
  apiKey: process.env.APICLAW_API_KEY,
});

const response = await client.chat.completions.create({
  model: "claude-sonnet-5-5",
  messages: [{ role: "user", content: "Explain request-based pricing in one sentence." }],
});
console.log(response.choices[0].message.content);

curl example

curl https://apiclaw.biz/v1/chat/completions \
  -H "Authorization: Bearer sk-your-apiclaw-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5-5",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Change the model field to switch families, for example to gpt-6-sol, deepseek-v4-flash or qwen3.8-max. The one API key for every model guide lists the available IDs.

Where flat-rate pricing fits

Flat-rate pricing suits workloads where the number of calls is steady but the size of each call is large or unpredictable. Coding agents are the clearest case, since they resend the full conversation on each turn. Internal tools that summarize long documents, review pull requests or run nightly batch jobs over big files fit as well, because the cost does not change when a document is longer than expected.

It suits workloads that stay below the daily quota. A product that serves thousands of end users from one key should size the plan against peak daily requests, and may need a custom plan above 8,000 requests a day.

It fits poorly where traffic is tiny or irregular. If you make a handful of short calls a week, a monthly fee buys capacity you will not use, and token billing will cost less.

Client compatibility and long context

Because the endpoint follows the OpenAI format, your client code stays the same: the same request shape, the same SDK calls and the same response objects. Check the model hub and the setup guides for the details of a specific model or tool before you depend on a particular feature.

Long prompts are part of the point. Models on the platform have up to 1M-token context, and APIClaw does not limit tokens per request beyond the model's own limit. A single call that sends a large codebase uses the same quota as a one-line question, apart from the model's weight. Pick the model by capability and weight, not by how long your prompt is.

Migrating from another OpenAI-compatible provider

If your code already uses an OpenAI-style client, migration is three edits: the base URL, the API key and the model name. Keep the key in an environment variable so the next switch is also a config change. Run your existing test prompts against a weight-1 model first, then move the real workload.

Then compare the invoice. Take your last month of token charges from the old provider and put them next to the plan price that matches your daily request count. The comparison pages show how APIClaw lines up against other routes, and the free trial lets you test before paying.

Which base URL to use

OpenAI-style clients, including Chat Completions and Responses, use https://apiclaw.biz/v1. Clients that speak the Anthropic Messages format, such as Claude Code, use https://apiclaw.biz with no /v1. Cursor agent mode uses https://apiclaw.biz/cursor. Setup guides are in the documentation, including Claude Code and Cursor.

What to check before you rely on any gateway

Look at the failure mode as well as the price. With a flat plan, the failure is hitting the daily quota. Ask what happens at the limit, when it resets and what a bigger tier costs. APIClaw refuses further requests until 00:00 UTC, and you can upgrade a tier or ask for a custom plan.

Check the payment path too. APIClaw takes crypto through NOWPayments and needs no card; see paying for an AI API with crypto. It is an independent gateway, not affiliated with Anthropic, OpenAI, Google or Cursor, and it routes requests to models available through publicly available sources and provider APIs. For help weighing several options, read the comparison pages and the guide to the best AI API gateway for coding agents.

Try it

The free trial includes 50 requests with no card. Create an account, copy your key and run one of the snippets above. If the response comes back, any OpenAI-compatible tool in your stack can use the same two settings.