Priced like the models you already pay for.
Predictable plans when you want a monthly run quota, or pure usage when you'd rather meter — the same ladder you already know from your model subscriptions, built for teams that ship.
beta
Open beta: every account runs on a generous free allowance — no card required. BYOK (your own keys) is unmetered. Card billing activates plans at GA; early accounts keep beta pricing.
Plans: runs are the headline. Everything else follows.
Every plan sets a monthly quota of full-mixture runs, a matching Moamao Cloud token credit, and a compute tier: the biggest panel your requests can summon.
Read a run as a decision, not a message: a second opinion on the answer you can't afford to get wrong, like the migration, the contract, or the architecture call. Scale is $20 for 25 runs, about $1 per decision; Max is $79 for 100 runs, about $1 per decision. Keep your daily-driver subscription for everything else.
Free
Taste real mixtures before you commit.
$0/mo
3 mixture runs / mo- Up to 2-model mixtures
- $2 Moamao Cloud credit / mo
- Hard 3-run cap — no overage
- 7-day trace retention
- Free in-browser demo, no account needed
- Community support
Scale
most popularFor builders shipping mixtures into real work.
$20/mo
25 mixture runs / mo- Up to 3-model mixtures
- $20 Moamao Cloud credit / mo
- Overage then $1.00 / run
- 90-day trace retention + zero-retention mode
- Priority queue
- Email support
Max
Full four-model mixtures at serious volume.
$79/mo
100 mixture runs / mo- Up to 4-model mixtures
- $79 Moamao Cloud credit / mo
- Overage then $0.75 / run
- 90-day trace retention + zero-retention mode
- Highest priority queue
- SSO coming
Enterprise
Custom volume, security review, and support.
Custom
custom volume- Up to 4-model mixtures
- Volume Moamao Cloud credit
- Volume run pricing
- 90-day trace retention + zero-retention mode
- SSO + audit logs
- Security review + DPA
- Dedicated support
Moamao Connector
A second frontier model inside Claude Code — no ChatGPT subscription needed. It runs on our compute, so unlike the free /moa plugin — which uses your own Claude and Codex subscriptions — you need no ChatGPT or Codex account of your own.
- Runs on our compute, so it needs no ChatGPT or Codex account of your own.
- codex_consult and codex_review are 1 call; moa_run is 2 (it runs two models).
- Failed calls are never charged.
claude mcp add --transport http moamao https://www.moamao.com/api/mcp5 free calls every month, then pay only for what you use. No monthly minimum.
Get started freeWhich billing is me?
Three ways in — pick by what you already have, not by plan size.
I have provider API keys
BYOK — billed per run. Your own provider rates, our orchestration on top.
"compute": "byok"I don't want to manage providers
Moamao Cloud — billed per token. Zero setup, no keys to wire up.
"compute": "moamao-cloud"I want it predictable
A plan bundles a monthly run quota plus a matching Moamao Cloud credit.
Scale · runs + creditThe BYOK bonus.
Most of our users bring their own keys — so we reward it. Full BYOK? Turn your Cloud credit into more runs, with a 20% bonus for not using managed compute. Any managed-compute budget you don't spend comes back to you as extra full-mixture runs — better than face value, every month.
2530
25 runs of credit → 30 BYOK runs
100120
100 runs of credit → 120 BYOK runs
The bonus applies to Cloud credit you leave on the table, so you only benefit when you're genuinely running on your own keys — exactly the users who don't need us to run the compute.
Prefer pure usage? No plan required.
Both compute modes work with no subscription at all: Moamao Cloud meters tokens, BYOK meters runs. Plans exist for included runs, matching credit, and bigger panels — not as a gate.
Moamao Cloud — per token
"compute": "moamao-cloud"Moamao Cloud is our managed mixture endpoint. It runs a full propose–critique–aggregate mixture on managed capacity with zero provider setup — you never manage a provider. Billed per token (input + output, across the whole mixture).
$2.00
per million tokens
One blended published rate across the whole mixture — input plus output. No per-model breakdown; the endpoint is opaque by design.
Moamao Cloud runs on managed capacity. When demand is high it may queue briefly — Build shows live status.
$2.00 / M tokens · zero provider setupBYOK — per run
"compute": "byok"Your own provider keys do the model work at your rates; Moamao charges an orchestration fee per run. With no plan at all, that fee is $0.02 per run, pay-as-you-go. On a plan, your monthly run quota (and per-run overage) covers these orchestration fees; Moamao Cloud is billed separately, metering tokens against your plan credit.
Heads up: a Claude or ChatGPT subscription isn't an API key — BYOK needs provider API keys from their consoles. No keys? Moamao Cloud needs none.
$0.02 / run pay-as-you-go · your provider rates applyOn a plan? Runs and tokens past your monthly quota continue automatically at your plan rates — with spend caps in the dashboard when you want a hard stop instead.
How usage is counted.
A run is one full mixture
On BYOK, a full mixture — proposals, critique, and judged synthesis — is one run, whatever the panel size or layer count. Model tokens stay on your provider bills; Moamao meters the run.
Moamao Cloud meters tokens
The managed mixture endpoint meters every token it processes — input plus output across the whole mixture — at one blended rate of $2.00 per million tokens, drawn from your plan credit.
Routed = ¼ run, failed = 0
When smart routing sends a prompt to a single fast model, it counts as a quarter of a run. A run that errors before returning an answer isn't billed at all.
Pricing questions: the fine print, up front.
A run is one full-mixture answer, and a full mixture is compute-heavy on purpose: several frontier models draft, critique each other, and a judge synthesizes the result. You're paying for up to four models' work per answer, not one API call — so a handful of runs a month buys a lot more thinking than the number suggests. We'd rather price the real work honestly than headline a big number that hides it.
Open beta: stop betting everything on one model.
Create a free account, generate an API key, and run your first mixture in minutes — or connect a repo in Build and watch it work. BYOK from day one.