unbiased ai

Onyx, citing your docs with Pareto

Onyx already knows where your company keeps things: it searches the Drive folders, Confluence spaces and other apps you connect, then writes an answer with the sources attached. Let Pareto write it: an admin adds Unbiased once as a custom provider, types pareto into the model list and makes it the default. Each reply comes from frontier and open models that Pareto selects and blends into one answer.

Choose Personal · $10/month See the setup →

Settings from Onyx’s docs50M tokens a week

An illustration of Onyx on Pareto. Not a screenshot.

Add Pareto as a custom provider

Onyx sends model calls through LiteLLM, and LiteLLM’s openai provider speaks the OpenAI Chat Completions format that Unbiased accepts. An admin sets it up once under Admin Panel > Language Models; the documents, connectors and agents your team built stay as they are.

  1. Get an Unbiased API key

    Choose a plan, then create a key in the Unbiased platform. New accounts are reviewed and approved by a person before the key works.

  2. Open Language Models

    Open Admin Panel at the bottom of the left sidebar and choose Language Models. Under Add Provider, find the Custom Models card marked LiteLLM-compatible providers and click Set Up.

  3. Fill in the provider

    Set Provider to openai, the LiteLLM provider ID for OpenAI-compatible endpoints. Paste your key into API Key and set API Base URL to https://api.unbiased.ai/v1, keeping the /v1. Leave API Version and Environment Variables empty, and give it a Display Name such as Unbiased.

  4. Type pareto by hand

    Under Models, click Add Model and enter pareto as the Model Name, with Input Type left on Text Only. Onyx will not save the provider until at least one model is listed.

  5. Choose who gets it, then make it the default

    Under Models Access, pick All Users & Agents, or Named Groups & Agents to limit it. Click Connect, then choose Pareto under Default Model on the Language Models page so new chats use it.

Settings follow Onyx’s own documentation: Onyx docs: Custom Inference Provider.

Custom Models (LiteLLM-compatible providers)
Provider:      openai
API Key:       your-unbiased-api-key
API Base URL:  https://api.unbiased.ai/v1
API Version:   (leave empty)
Display Name:  Unbiased

# Models > Add Model
Model Name:    pareto
Display Name:  Pareto
Input Type:    Text Only

Onyx also has a second Custom Models card, for OpenAI-compatible endpoints. That one fills its model list by asking the server for /models, which the Unbiased API does not serve, so use the LiteLLM card, where you type the model name yourself.

Your sources stay. The writer changes.

Onyx still does the searching and the citing. Pareto writes the answer from what Onyx finds.

One model string, several models behind it

Pareto is Unbiased’s blended AI model. On each request it selects frontier and open models and blends their outputs into one answer, so Onyx talks to one model string, pareto, and never has to pick.

“When the frontier improves, Pareto improves; when frontier prices fall, you should benefit.” From the pricing page.

How a Pareto answer gets made →
An illustration, not a wiring diagram. The blend includes frontier and open models, and its composition can change.

What an agent task costs on Pareto

One Onyx answer can take several model calls: decide what to search, read the passages, then write. The model card publishes cost and success rate for Pareto on two public agent benchmarks.

Terminal-Bench 4.0Tasks an agent completes in a terminal
$0.48/task
50.8% solved
DeepSWE v1.1Changes an agent makes in a real codebase
$0.24/task
69.9% solved

Pareto 26.10, preliminary October 1, 2026 runs that may change; denominators and the cost method still require confirmation. Cost is the reported mean USD per task at API rates, not an estimate for your workload. Model card and method.

Pay for the answers by the month

Onyx decides what to read. Pareto is the part you pay us for: pick a monthly allowance, or pay per token on Team. The same key works in any other tool you point at Pareto.

Personal

$10/month

  • 50M tokens a week
  • 20 requests per minute
  • For one person
Choose Personal

Personal Max

$100/month

  • 600M tokens a week
  • 60 requests per minute
  • For one person
Choose Personal Max

Team

Pay as you go

  • Prepaid credits at published rates
  • 300 requests per minute
  • No weekly token cap
Get started with Team

Subscriptions are for one person and are not unlimited. New API accounts are reviewed by a person before access turns on, and commercial or production use needs a separate written agreement. Rates and limits are on Pricing.

Onyx on Pareto, questions answered

Which provider do I choose in Onyx?

The Custom Models card labeled LiteLLM-compatible providers, with Provider set to openai and API Base URL set to https://api.unbiased.ai/v1. Onyx’s docs show vertex_ai as an example provider; that one is for Google, not for Unbiased. Format compatibility is not a promise that every Onyx feature works, so tell us if one doesn’t.

Why can’t I find pareto in the model list?

Unbiased does not publish a model list, so Onyx cannot fetch one. In the LiteLLM form, click Add Model under Models and type pareto. If an admin can see it but your colleagues can’t, open the provider again and check that Models Access includes their groups, or set it to All Users & Agents.

Does Pareto see all of our documents?

No. Onyx searches the documents it has indexed, then sends the question and the passages it picked to the model. Pareto answers from what it receives. Your documents and the index stay in Onyx.

Does each person need their own Unbiased key?

No. The key lives on the provider the admin sets up, and everyone given access uses it. Personal plans are for one person, so an Onyx instance shared by a team belongs on the Team plan. Running Onyx for a company in production also needs a separate written agreement with Unbiased; platform teams can start here.

Can I keep our other models?

Yes. Pareto is one provider among however many you have, and users can still pick another in the model selector. To switch back, set a different Default Model on the Language Models page.

Is Pareto made by the Onyx team?

No. Pareto is Unbiased’s model. Onyx is made by Onyx (formerly Danswer), and Unbiased is not affiliated with or endorsed by them. You keep Onyx and point it at the Unbiased API.

What happens when I hit the weekly allowance?

Requests over your plan’s limits return an HTTP 429 error. Personal includes 50M tokens a week and 20 requests per minute. Personal Max raises that to 600M a week and 60 requests per minute for $100/month. Team uses prepaid pay-as-you-go credits with no weekly token cap.

Do I choose between models?

No. You send every request to one model string, pareto. Pareto selects frontier and open models for each request and blends their outputs into one answer.

Is my code kept?

Personal and Personal Max are not our zero data retention tier: prompts and responses may be kept up to 30 days, as section 5(c) of the Terms explains. For work that must not be kept, ask about the Zero Data Retention API on the Private AI page.

Pareto works in other tools too

The same key and the same model string go into each of these. See them all.

Product names belong to their owners. Except where a guide says otherwise, Unbiased is not affiliated with or endorsed by the makers of the tools on this page.

Ask Onyx the question everyone asks HR

One provider, one model name and the sources you already connected. Pareto starts at $10 a month for 50M tokens a week.

Choose Personal · $10/month See the setup →