Cognitivers docs

Cognitivers serves open-weight models over an API that speaks OpenAI chat completions. You keep the client you already use, change two values, and you are done: the base URL and the model string.

curl https://api.cognitivers.com/v1/chat/completions \
  -H "Authorization: Bearer sk-cog-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cog-fast",
    "messages": [{ "role": "user", "content": "Summarise this incident timeline." }]
  }'

Start here

Quickstart

One request in three lines, in cURL, Python and Node.

Models and prices

What is served, how much context it takes, and what it costs.

Tools

Setup for the coding agents and editors you already run.

The request reaches the weights exactly as you wrote it, and what comes back is the model's answer rather than a filter's. There is no classifier scoring your prompt before inference, nothing rewriting it on the way in, and no moderation layer between the model and you. Your system prompt is the only policy in the loop.

That matters for work where a refusal halfway through is a failed run: security automation, OSINT and journalism, moderation datasets, harm reduction, long-form fiction, and agents that are not allowed to stall.

Compatibility

If your client speaks OpenAI chat completions, it already speaks Cognitivers. Change the base URL and the model string, and keep everything else.

Where to go next

New to the API: Quickstart, then Authentication.
Bringing your own tool: Tools, which covers seventeen of them with verified configuration.
Deciding whether you can use it: Acceptable use and Privacy.
Working out the bill: Pricing and Plans.