unbiased ai

GitHub Copilot CLI, with your own key and Pareto

Copilot CLI has a bring-your-own-key mode, and it asks for very little: a base URL, an API key and a model name. Give it Unbiased’s, and the agent that reads your files, edits them and runs your tests does all of it on Pareto. Pareto selects frontier and open models for each request and blends their outputs into one answer.

Choose Personal · $10/month See the setup →

Tested live with GitHub Copilot CLI 1.0.10150M tokens a week

A demonstration of GitHub Copilot CLI on Pareto, recreated from a real run. Not a screenshot.

Set it up with three variables

Copilot CLI reads its BYOK settings from the environment. Export them in your shell, or add them to your shell profile so every copilot session starts on Pareto.

  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. Point Copilot CLI at Unbiased

    Set COPILOT_PROVIDER_BASE_URL to https://api.unbiased.ai/v1 and put your key in COPILOT_PROVIDER_API_KEY. Keep the /v1: the default openai provider type adds /chat/completions after it.

  3. Name the model pareto

    Set COPILOT_MODEL to pareto, or pass --model pareto. Unbiased does not serve a model list, so type the name yourself.

  4. Start Copilot CLI

    Run copilot in your project. It asks before it edits a file or runs a command. For a one-shot run, use copilot -p "…" with --allow-all-tools (or --allow-tool for specific tools), since prompt mode can’t stop to ask.

Tested on October 20, 2026GitHub Copilot CLI 1.0.101 on Pareto (pareto). It found the off-by-one in a failing Python test, edited the file and reran the test to green, with no GitHub sign-in. Run record.
Shell
export COPILOT_PROVIDER_BASE_URL=https://api.unbiased.ai/v1
export COPILOT_PROVIDER_API_KEY=your-unbiased-api-key
export COPILOT_MODEL=pareto

# interactive
copilot

# one-shot
copilot -p "Fix the failing test in test_chunk.py" --allow-all-tools

COPILOT_PROVIDER_TYPE defaults to openai and COPILOT_PROVIDER_WIRE_API to completions. Those defaults are what we tested, so leave both unset.

GitHub Copilot CLI stays. The model changes.

The agent, its approvals and your scripts stay put. The provider, the model name and who sends the bill are what change.

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 GitHub Copilot CLI 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

Agents make many model calls per task, so the price of a finished task matters more than the price of a token. The model card publishes both 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 GitHub Copilot CLI by the month

One API key works across GitHub Copilot CLI and any other tool you point at Pareto. Pick a monthly allowance, or pay per token on Team.

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.

GitHub Copilot CLI on Pareto, questions answered

Do I need a GitHub account to use Copilot CLI with Pareto?

No. GitHub’s authentication docs say that with your own provider keys (BYOK), Copilot CLI connects directly to the provider without a GitHub account or token. Our test of version 1.0.101 on October 20, 2026 never signed in: it read a failing Python test, fixed the bug and reran the test to green. Without a sign-in, /delegate, the GitHub MCP server and GitHub Code Search are unavailable. You can sign in with /login and keep Pareto as the model.

Which provider type and wire API should I choose?

Leave both at their defaults: openai and completions. That sends OpenAI Chat Completions requests to https://api.unbiased.ai/v1, the path we tested. Unbiased also accepts Anthropic Messages and OpenAI Responses requests, but we have not tested Copilot CLI’s anthropic type or responses wire API. Format compatibility is not a promise that every feature works, so tell us if one doesn’t.

Does Pareto meet Copilot CLI’s model requirements?

Yes for the path we tested: Copilot CLI 1.0.101 streamed and called tools against Pareto in our run. GitHub’s docs say a BYOK model must support tool calling and streaming, and recommend a context window of at least 128K tokens. Pareto has a context window of about 1M tokens. If you need to cap token counts, Copilot CLI has COPILOT_PROVIDER_MAX_PROMPT_TOKENS and COPILOT_PROVIDER_MAX_OUTPUT_TOKENS; our test needed neither.

Does anything still go to GitHub?

Your prompts and code context go to api.unbiased.ai. GitHub’s docs say that even with BYOK and no GitHub sign-in, Copilot CLI still sends telemetry as usual unless you use offline mode. Offline mode is meant for local model providers, so it does not fit a hosted model like Pareto.

Can I switch back to GitHub’s models?

Yes. Unset COPILOT_PROVIDER_BASE_URL, COPILOT_PROVIDER_API_KEY and COPILOT_MODEL, then sign in to GitHub if you haven’t, and Copilot CLI uses the models that come with your Copilot plan again.

Is Pareto made by the GitHub Copilot CLI team?

No. Pareto is Unbiased’s model. GitHub Copilot CLI is made by GitHub, and Unbiased is not affiliated with or endorsed by them. You keep GitHub Copilot CLI 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.

Hand Copilot CLI the test that keeps failing

Three variables, one key and the same copilot command. Pareto starts at $10 a month for 50M tokens a week. Compare it with other coding plans.

Choose Personal · $10/month See the setup →