Nossic's own AI, run in the UK

One front door to our AI. Your request waits its turn, it never fails for being busy.

Nossic AI Gateway puts Nossic's AI behind one OpenAI-compatible API. Every app gets its own key and allowance, every request queues fairly until the AI is free, and everything runs on Nossic's servers in the UK.

Open the app See the API

Public sign-ups Coming soon For now it's open to Nossic's own apps and invited accounts.

A busy moment, as your app sees it

POST /v1/chat/completions   "stream": true

: waiting, position 3
: waiting, position 1
data: {"choices":[{"delta":{"content":"Hello"}}]}
data: {"choices":[{"delta":{"content":"! How can"}}]}
data: {"choices":[{"delta":{"content":" I help?"}}]}
data: [DONE]

How it works

Four steps, and your app is talking to Nossic's AI.

1
Get a key
Sign in with your Nossic ID and make an AI access key for your app. Each app gets its own, so you can see what each uses.
2
Point your app at it
Anything that speaks the OpenAI API works: change the base URL to app.ai.nossic.uk/v1 and use your key.
3
It waits its turn
When the AI is busy, your request queues fairly and starts as soon as it's your go. No "too busy" errors to retry.
4
Watch your usage
See tokens in and out by day, key and model, and how much of your allowance is left before it resets.

Built for apps that rely on AI

A fair queue, with no time limit

Requests wait until the AI is free, however long that takes, and owners take turns so one busy app can't hold up everyone else. Give up, and your request leaves the queue.

OpenAI-compatible

Chat with streaming, tool calling, images and JSON mode; completions; embeddings; and the models list. Your existing SDKs and tools just work.

A key for every app

Name each key, limit it to certain models or addresses, give it its own token limit or an expiry date, and revoke it in one click.

Allowances you can see

Token allowances that reset every few hours, days or each month, with exact counts from the AI itself and a usage chart for every key and model.

For you or your team

Use it personally, or for an organisation from Nossic Accounts: everyone in it sees the usage, and its owners and admins manage the keys.

Private by design

Only counts and timings are kept, never your prompts or replies. The AI runs on Nossic's own hardware in the UK; nothing is sent to an outside AI company.

Two settings, and you're in

Use the OpenAI library you already have. Change the base URL and the key; turn off the time-out, because a busy moment means waiting, not failing.

Base URL: https://app.ai.nossic.uk/v1
Key: an AI access key from the app (it starts nai_)
Streaming shows your place in the queue while you wait.
Embeddings for search, from the same key.

Python

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://app.ai.nossic.uk/v1",
    api_key=os.environ["NOSSIC_AI_KEY"],
    timeout=None,  # wait as long as it takes
)
reply = client.chat.completions.create(
    model="nossic-smart",
    messages=[{"role": "user", "content": "Hello!"}],
)

Models

Ask for a nossic- name: Nossic can move it to a better model later, and your app keeps working without a change.

nossic-smart

Our best model

Chat, tool calling, images and long documents, with 256k tokens of context. Today: Ornith 1.5 35B.

nossic-fast

Quick answers

For live chat and short replies. Today it's the same model as nossic-smart; a smaller, faster one joins when we add more AI hardware.

nossic-embed

Embeddings for search

Turns text into vectors for search and matching (nomic-embed-text, 768 dimensions).

Ready to plug in?

Open Nossic AI Gateway