WETWARE A Plystra experiment.

WETWARE / PUBLIC ALPHA

OpenAI-compatible inference, powered by actual humans.

Put a person behind a model ID and call them with the ordinary OpenAI SDK. When they are available, they answer the request themselves.

Create a model Read the API docs
Python 
from openai import OpenAI

client = OpenAI(
    api_key="sk-live-example",
    base_url="https://api.wetware.example/v1",
)

response = client.chat.completions.create(
    model="moses",
    messages=[
        {"role": "user", "content": "hello"}
    ],
)

Compatibility

Change three fields.

An existing OpenAI client needs a different base URL, API key, and model ID. The rest of the interface stays familiar.

base_url
https://api.wetware.example/v1
api_key
sk-live-…
model
moses

The runtime

A person occupies the model slot.

  1. Application
  2. Wetware API
  3. model="moses"
  4. Human reply
  5. API response

Availability

Online means callable.

A model accepts requests only while an authenticated operator session is live. Offline models return 503 immediately. Requests are never held for later.

Wetware models are operated by people. Protocol compatibility does not imply artificial intelligence.

Inference backend: Homo sapiens · Parallelism: 1