AI providers · OpenAI · updated September 2, 2026

Run your AI employees on OpenAI

Bring your own OpenAI key and Coordinator's agents run on GPT models — or point the base URL at any OpenAI-compatible endpoint you operate, billed to you directly at zero markup.

Free. No card. About two minutes to set up.

What you can delegate

A OpenAI-powered employee, not a chatbot

An agent here holds roles exactly like a person does, and its ceiling is its roles' ceiling: whatever someone with those roles could do in Coordinator, you can hand over. OpenAI decides how well the work gets done — not how far it reaches.

Money & books
  • Read a forwarded vendor PDF into a draft bill, validated before it touches the books
  • Watch receivables aging and draft the chase
  • Answer P&L and client-profitability questions from the live ledger
Projects & work
  • Create, file and update activities across your pipelines and boards
  • Work a queue — claim the next item, do it, hand off
  • Run the check-in round-up and keep stages honest
Time & billing
  • Log time as it works
  • Sweep unbilled hours into a draft invoice
  • Keep the time ledger clean enough to bill from
Records & forms
  • Triage form submissions into records and the right pipeline
  • Keep client records current as work happens
  • Find anything — records, contacts, history — on request
Scheduling
  • Prep today's meetings from the calendar
  • Tend booking-link intake and meeting polls
  • Keep resources and rooms straight
Chat & threads
  • Draft replies, follow-ups and summaries inside your team's threads
  • Announce what needs announcing
  • Turn a meeting thread into the follow-up tasks
The short list it can never do

Hold the Owner role. Message outside your organization before you deliberately climb the outbound ladder. Appear human in the audit trail — agent work is labeled agent work, always. Keep going after you hit pause. Everything else opens closed and is yours to grant, role by role.

The full AI-employees story — roles, fences, the outbound ladder
Why OpenAI

What OpenAI itself brings

Any provider on this site can power the delegation above. These are the reasons to pick OpenAI in particular.

  • Native PDF input on the current lineOpenAI is one of three providers here that takes a PDF as a file — and GPT-5.6 Luna reads a one-page receipt for a fraction of a cent, on the current lineup rather than a legacy one.
  • The escape hatch is built inThis connection deliberately accepts a custom base URL. Azure-style gateways, LiteLLM, a local proxy, an on-prem model server — if it speaks OpenAI's wire protocol, your agents can run on it. Infrastructure you control, same agents.
  • ChatGPT can drive Coordinator tooSeparately from your API key, Coordinator's MCP server lets ChatGPT operate your workspace with scoped, expiring, reversible-writes-only tokens. Key in, MCP out — your choice.
  • Models

    OpenAI models your agents can use

    Prices and lineup verified September 2, 2026. In the product, the picker is fetched live from OpenAI on your own key — it never goes stale, and a custom model id always works if it validates.

    GPT-5.6 Lunacheap + fast

    $0.20/$1.20 per million tokens — the current budget tier, and the natural extraction default

    GPT-5.6 Terrabalanced

    $2/$12 — the mid-tier of OpenAI's current flagship line

    GPT-5.6 Solflagship

    $4/$20 (promotional into late November 2026) — the top of the current line

    GPT-4o familylegacy

    still served on the API, but OpenAI is winding the era down — the o-series and dated snapshots start disappearing October–December 2026, so pin new work to 5.6

    Anything behind your gatewaybase URL

    point the base URL at Azure-style gateways, LiteLLM or an on-prem server and its catalogue is the list

    Media through Coordinator's pipeline: PDFs and images. The authority on current rates is OpenAI's pricing page.

    What developers say · September 2, 2026

    OpenAI's API, as its users describe it

    Recurring themes from public developer forums and reviews — the likes and the complaints — and how each plays out when the provider runs inside your own workspace.

    What they like
    • The developer platform's maturity is the consistent draw: polished SDKs, batch and flex processing at half rate, and measured uptime around 99.9% through mid-2026.
    • The current GPT-5.6 line (Sol, Terra, Luna) is built for agent workloads with long context across all three tiers — and even the price critics acknowledge the capability arc from 5.5 onward.
    • GPT-5.6 Luna at $0.20/$1.20 per million tokens is a genuinely excellent cheap tier — a one-page receipt reads for a fraction of a cent on the current line, not a legacy one.
    What they complain about — and how it plays here

    2026 was the year of forced migrations: the Assistants API was hard-shut in August with no automated migration, and dated model snapshots teams had pinned (as recommended) get removed on a treadmill.

    Here: Coordinator deliberately speaks the boring wire protocol — plain chat completions — not the stateful platform APIs that got shut down. The Assistants sunset didn't touch a single Coordinator agent, and that's by design.

    Pricing whiplash: the flagship line doubled with GPT-5.5 in April 2026 ($5/$30), and the current 5.6 Sol’s $4/$20 is a promotional rate with an end date — forum sentiment on the trendline is blunt.

    Here: Nothing about your agents is built on OpenAI. If the price stops making sense for a role, point that agent at another provider — or a cheaper OpenAI model — without rebuilding anything.

    Identity verification is now required for the newest models, and developers outside the US report failed verifications and lost simplicity.

    Here: Your key's actual reach is tested at connect time, so a verification-gated model fails loudly in settings rather than silently in an agent's first task. And the base-URL override means a gateway you control works exactly like the real endpoint.

    Ours to get right

    How Coordinator treats your OpenAI key

    The section above is OpenAI's report card. This one is ours — the plumbing decisions that make handing an agent your API key boring instead of brave.

    Stored encrypted, shown as last-four

    Your key is encrypted like every credential in Coordinator, and after connecting, nothing ever displays more than its last four characters.

    Proven before it's stored

    Connecting is a live round-trip to OpenAI — a key that doesn't actually work can't be saved, so failures happen in settings, not mid-task.

    Billed by OpenAI, never by us

    Zero markup, no metering, no shared key anywhere in the system. Every agent turn bills your own account at the vendor's own prices.

    Gateway-honest validation

    Because this connection also serves gateways whose catalogues differ, validation checks your key against the model list it actually exposes — a proxy setup validates the same way the real endpoint does.

    Getting started

    From no key to a working agent in a few minutes

    Create a project key at platform.openai.com → API keys, and give it model read + write permissions.

    First, at platform.openai.com
    1. 1

      Sign in at platform.openai.com (this is the developer platform — a ChatGPT Plus subscription does not include API credit).

    2. 2

      Pick the project you want the spend to land in, top right.

    3. 3

      Open API keys in the left sidebar and choose Create new secret key.

    4. 4

      Set Permissions to All, or if you use restricted keys, grant Model capabilities → Write and Models → Read.

    5. 5

      Copy the key immediately — it is shown once.

    6. 6

      Under Settings → Billing, add a payment method or credits. New projects start with none and every call returns an insufficient-quota error.

    Then, in Coordinator — the actual screens
    Connect OpenAI-compatible
    API key
    sk-········ab12
    Validate & save

    Key verified — a live call to OpenAI confirmed it before anything was stored. Only the last four characters are kept visible.

    Settings → AI: pick OpenAI-compatible, paste the key. A live call to OpenAI verifies it before anything is stored.

    Default model
    GPT-5.6 Luna
    GPT-5.6 Lunacheap + fast
    GPT-5.6 Terrabalanced
    GPT-5.6 Solflagship
    GPT-4o familylegacy

    Fetched live from your key — never a stale list. Custom model ids work too.

    Choose a default model from your key's live catalogue — or set one per agent later.

    What it costs

    Billed per token on your own account. gpt-5.4-mini is the cheap default and reads a typical one-page receipt for a fraction of a US cent. Larger models cost proportionally more; check the pricing page for current rates.

    Watch out

    This endpoint takes images only. A PDF bill will not extract here — connect Anthropic or Gemini for PDFs, or attach a photo/screenshot instead.

    PDFs go in natively

    Coordinator sends OpenAI your PDFs as its documented file content part — the vendor's own PDF path, no rasterising. One caveat if you use the base-URL override: your gateway has to support OpenAI's file part too, or the PDF leg fails with the gateway's own error. Images are unaffected either way.

    Your key, your spend

    OpenAI (or whoever operates your compatible endpoint) bills you directly. No markup, no shared key.

    FAQ

    OpenAI + Coordinator, asked directly

    Can I use Azure OpenAI or a local model server?

    Yes — the OpenAI connection accepts a base-URL override, so any OpenAI-compatible endpoint works: Azure-style gateways, LiteLLM, a local proxy, an on-prem server.

    Which models are supported?

    Whatever chat models your key or gateway exposes — the current GPT-5.6 line (Sol, Terra, Luna), the still-served legacy families, or anything behind a compatible gateway. The picker lists your live catalogue and custom ids save if they validate.

    Do I pay Coordinator for OpenAI usage?

    No. Coordinator is free; your provider bills your key directly.

    Not sure OpenAI is the one?

    Swap the brain, keep the team

    Your agents aren't built out of OpenAI — they're built out of roles, permissions and workflows. The model behind an agent is a setting: change it in a minute and the team, the history and the fences all stay put. Every provider we support gets the same treatment, and the same honest page:

    100% free — every feature, every seat

    Put OpenAI on the payroll

    Coordinator is free — the only bill is the one OpenAI already sends you.

    Free. No card. About two minutes to set up.