Chat and completions

The OpenAI-compatible door. If your code already talks to OpenAI, change two things — the base URL and the key — and it talks to us.

Calling it

curl https://api.sociaro.com/v1/chat/completions \
  -H "Authorization: Bearer $SOCIARO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "model": "anthropic/claude-sonnet-5",
        "messages": [{"role": "user", "content": "Explain a hash join in two sentences."}]
      }'

With the OpenAI Python library:

from openai import OpenAI

client = OpenAI(api_key="sk-...", base_url="https://api.sociaro.com/v1")

reply = client.chat.completions.create(
    model="anthropic/claude-sonnet-5",
    messages=[{"role": "user", "content": "Explain a hash join in two sentences."}],
)
print(reply.choices[0].message.content)

Streaming

Set "stream": true and read server-sent events, exactly as you would from OpenAI. Long generations are safer streamed: a connection that produces nothing for a long time can be cut by the network between us, and a stream keeps it alive by having something to say.

Which models

GET /v1/models lists what your key may reach. The catalogue has every model with its parameters and what it costs.

Names read as vendor, then a slash, then the model — and the vendor is the one who made it, not whoever we happen to buy it through.

What we pass through

Standard parameters — temperature, top_p, max_tokens, tools, response_format — go to the model unchanged. A parameter a model does not support is refused rather than silently dropped, so a request that returns is a request that was understood.