中文

Test an API gateway

Check model responses, protocol compatibility, latency and billing evidence. Tests run through an independent service. Use a temporary API key; the BiWu.Ai website does not store your raw key.

Chat Completions / Responses / Claude / Gemini / GrokKeys go only to the test service

Connection settings

Enter the gateway endpoint and a temporary API key, then choose a model. Sign in to start a test.

Local test service readyNo tests yet
Create a separate key with a small allowance. Avoid production or high-balance keys, and revoke it after testing. The raw key is sent only to the independent test service; it is not saved in the report or by the BiWu.Ai website.
Choosing a model selects a suitable protocol. You can still edit the model name.
Estimates use the model's official rate. Your gateway determines the actual charge.

Selected: OpenAI Responses · /v1/responses — for the OpenAI Responses API and agent tools · Treat the source as unknown and interpret the results cautiously.

Estimated cost: 9k token 输入 / 3.5k token 输出 / 5 次请求,约 $0.15 USD;OpenAI API GPT 5.5 标准价。

Test results

Each test appears below. Open its report when it's complete to review the evidence.

Not started

No test results yet

Choose a model above and start a test. Your newest results will appear first.

Use a baseline

Compare against an official or trusted service. One plausible answer is not enough.

Know the source

Official APIs, cloud providers, account pools and unofficial routes are labelled separately.

Repeat your checks

Mixed routes and account pools need repeated requests. Avoid drawing conclusions from one result.