AI
Mock the OpenAI API
Chat completions and embeddings that cost nothing and answer instantly.
Every test run against a real LLM API costs tokens and takes seconds. These routes return the same chat.completion and embedding envelopes with the same usage accounting, so the code around your model calls — parsing, retries, streaming fallbacks, error handling — can be developed and tested for free. A conditional route returns a 429 when you ask for it, which is the case most retry logic is never actually tested against.
What you get
/v1/chat/completionstemplated200 OK/v1/chat/completionsconditional429 Too Many Requests/v1/embeddingstemplated200 OK/v1/models200 OKResponse template
POST /v1/chat/completions
Values in double braces are rendered per request.
{
"id": "chatcmpl-{{ uuid4() }}",
"object": "chat.completion",
"created": "{{ random_int(1700000000, 1760000000) }}",
"model": "{{ request.body.model }}",
"choices": [
{
"index": 0,
"message": {
"role": "assistant",
"content": "This is a mocked completion from MockBase."
},
"logprobs": null,
"finish_reason": "stop"
}
],
"usage": {
"prompt_tokens": 19,
"completion_tokens": 12,
"total_tokens": 31
}
}Good for
- Develop against an LLM API without spending tokens on every reload
- Test rate-limit and retry handling deterministically
- Keep CI fast and offline, with no API key in the pipeline
Response payloads are representative examples using this API's real field names and object shapes, not an exhaustive replica of its schema. MockBase is an independent tool and is not affiliated with, endorsed by, or sponsored by OpenAI.