CognitivessAI offers three kinds of models. They all share one API key and the same OpenAI-compatible endpoints — you just pick a model by name in the model field.

Authentication & endpoints

Every request must include your API key in the Authorization header as a bearer token. Your key looks like:

ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIOx1xMtshDsD7HB0dQ6VRggUK8/a10ixPL5C4oLEucVM

You need an account to generate a key. Sign up free to get yours — it is shown only once, so store it securely.

The same key works for every model below. Endpoints (point base_url at https://api.cognitivess.com/v1):

• POST https://api.cognitivess.com/v1/chat/completions — OpenAI-compatible chat completions (stream & non-stream)
• POST https://api.cognitivess.com/v1/messages — Anthropic Messages API (stream & non-stream)
• POST https://api.cognitivess.com/v1/responses — OpenAI Responses API (stream & non-stream)
• GET https://api.cognitivess.com/v1/models — list available models

See the Models page for specs and pricing of each hosted model.

1 · Cognitivess-1 — our model

Our flagship model, developed in-house. Text in, text out. It is the default and the best choice for most text workloads — coding, reasoning, long-context chat.

# 1) Install the CLI (Linux / macOS / WSL — needs only Python 3)
curl -fsSL https://api.cognitivess.com/install | sh

# 2) Save your API key once (prompted securely, not echoed)
cognitivess login

# 3) Launch Claude Code under Cognitivess-1
cognitivess launch claude

Choose your language

from openai import OpenAI

client = OpenAI(
    api_key="<YOUR_API_KEY>",  # cheia ssh-ed25519 din dashboard
    base_url="https://api.cognitivess.com/v1"
)

response = client.chat.completions.create(
    model="Cognitivess-1",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello, how are you?"}
    ],
    max_tokens=131072,
    temperature=0.7
)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "<YOUR_API_KEY>",
  baseURL: "https://api.cognitivess.com/v1",
});

const response = await client.chat.completions.create({
  model: "Cognitivess-1",
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "Hello, how are you?" },
  ],
  max_tokens: 131072,
  temperature: 0.7,
});

console.log(response.choices[0].message.content);

curl https://api.cognitivess.com/v1/chat/completions \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Cognitivess-1",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, how are you?"}
    ],
    "max_tokens": 131072,
    "temperature": 0.7
  }'

Anthropic Messages API

All hosted models also speak the Anthropic Messages API — point the Anthropic SDK at us and pass your key as api_key (sent as x-api-key). Works for Cognitivess-1 and kimi-k3.

from anthropic import Anthropic

client = Anthropic(
    api_key="<YOUR_API_KEY>",  # cheia ssh-ed25519 din dashboard
    base_url="https://api.cognitivess.com",  # fara /v1; SDK-ul adauga el
)

message = client.messages.create(
    model="Cognitivess-1",
    max_tokens=131072,
    system="You are a helpful assistant.",
    messages=[{"role": "user", "content": "Hello, how are you?"}],
)

print(message.content[0].text)

2 · Cloud models — kimi-k3

Cloud models are third-party models we host and serve through the same API, with the same key. You call them exactly like Cognitivess-1 — just change the model field. Currently available: kimi-k3 (Moonshot AI Kimi K3).

Text request

from openai import OpenAI

client = OpenAI(api_key="<YOUR_API_KEY>", base_url="https://api.cognitivess.com/v1")

response = client.chat.completions.create(
    model="kimi-k3",
    messages=[{"role": "user", "content": "Explain quantum entanglement simply."}],
    max_tokens=512,
)

print(response.choices[0].message.content)

Image input

Pass multimodal content as an array of parts. Both image_url (public URL) and base64 data URIs are accepted.

from openai import OpenAI

client = OpenAI(api_key="<YOUR_API_KEY>", base_url="https://api.cognitivess.com/v1")

response = client.chat.completions.create(
    model="kimi-k3",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What's in this image?"},
            {"type": "image_url", "image_url": {"url": "https://example.com/photo.jpg"}},
        ],
    }],
    max_tokens=512,
)

print(response.choices[0].message.content)

Video input

from openai import OpenAI

client = OpenAI(api_key="<YOUR_API_KEY>", base_url="https://api.cognitivess.com/v1")

response = client.chat.completions.create(
    model="kimi-k3",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "Summarize what happens in this video."},
            {"type": "video_url", "video_url": {"url": "https://example.com/clip.mp4"}},
        ],
    }],
    max_tokens=1024,
)

print(response.choices[0].message.content)

For base64 instead of a URL, use a data URI: "url": "data:image/jpeg;base64,<BASE64>" (image) or "data:video/mp4;base64,<BASE64>" (video). Multimodal input works on both streaming and non-streaming requests.

3 · Local models — run any GGUF offline

Beyond the hosted models above, the cognitivess CLI can run any open GGUF model fully offline on your own machine — no API key, no network, your data never leaves your device. Useful for privacy-sensitive work, air-gapped machines, or zero-cost experimentation.

# Install the CLI (if not already)
curl -fsSL https://api.cognitivess.com/install | sh

# Download a local GGUF model (cached on your machine)
cognitivess pull qwen2.5-7b-instruct

# Run it in a browser web UI
cognitivess serve qwen2.5-7b-instruct

# ...or run it inside Claude Code, fully offline
cognitivess launch claude --local qwen2.5-7b-instruct

Quality is lower than the hosted cloud models and depends on your hardware, but it is free and private. See the CLI reference for the full list of commands, and Edge & Devices for on-device guidance.