Your account has been deleted. Your data has been purged, any subscriptions were cancelled, and this email address can sign up again as a fresh account.
One subscription · Every top model
Chat with the best models from every major lab — routed for you, and metered honestly in MultiO tokens at published rates. Every charge is shown on the message itself. Or ask once and compare them side by side.
Free trial — no card required. Then Pro is $20/month for 80,000,000 tokens.
Every message is charged in MultiO tokens at the model’s published rate — the exact charge, the real tokens used, and your remaining balance, shown on the message itself.
The new model sees the whole thread, your files and your memory. Hop from GPT to Claude to Gemini inside one conversation, freely. And your data is yours — export everything, anytime.
One subscription. Every top AI model. And your tokens go further: Auto sends each message to the most cost-efficient model that can handle it — quick questions to fast, low-rate models, hard ones to flagships. On a typical workload, a grant lasts ~3–4× longer with Auto than if every message went to a flagship model.
Median monthly usage on Pro with Auto routing — of the 80M-token grant
The same usage, always sent to flagship models
Repeat questions — answered from cache, free
Modelled on a 300-message/month median workload at published rates; your mix will vary. Every message shows its exact token charge on its receipt.
A taste of MultiO's compare view. The answers below are simulated (demo mode), but the metering is the real thing — this is exactly the receipt a real message carries.
Demo mode: simulated answers, metered at each model's published rate — the same maths as a real receipt. Sign up and your first 3,000,000 tokens are free.
$0
3,000,000 tokens
Small & mid models
$20/mo
80,000,000 tokens / month
All models, including flagship
$39/mo
160,000,000 tokens / month — one pool, up to 5 people
All models, including flagship
$25/seat/mo
80,000,000 tokens / month per seat — one shared pool for the organisation’s API keys
All models, including flagship · Org keys & BYOK · Admin controls
+$12/seat/mo
Each seat gets its own 80,000,000 tokens / month
All models, including flagship
Every plan includes the developer API — auto-routing, per-request receipts and per-key budget caps.
Token rates follow the providers' published list prices, converted at a fixed pivot: 1,000,000 MultiO tokens = $0.15 of standard-tier usage.
| Model | Tier | Input · tokens per 1,000 | Output · tokens per 1,000 |
|---|
No models match those filters.
Web search: 100,000 tokens per search, shown as its own receipt line.
Slide to your pace and see it in plain words — Auto routing against pinning every message to a flagship.
Modelled on a 300-message/month median workload at published rates; your mix will vary. Every message shows its exact token charge on its receipt.
The same models, the same metering, the same receipts — whether it’s just you, your code, or your whole team.
One subscription for every top AI model. Leave routing on Auto and each message goes to the most cost-efficient model that can handle it — or pin a flagship yourself. Every message shows its exact token charge on its receipt.
See plansAn OpenAI-compatible API on every plan — no separate account. Send model: "auto" for the same routing, get a receipt on every call (cache hits cost 0 tokens), and set hard daily and monthly budget caps per key, enforced before a call is made. 60 requests per minute per key.
Teams is $25 per seat/month: an organisation with owner, admin and member roles, a pooled allowance of 80,000,000 tokens per seat per month, organisation API keys that draw the pool, and org BYOK — calls on your own provider keys are not token-charged. Admins see per-member usage and an audit log; members’ personal chats and usage are never visible to org admins.
See Teams pricingSSO/SAML, SCIM, SOC 2 and custom terms are on our roadmap. Talk to us: hello@multio.ai
Enter your account email and we will send you a reset link.
You're almost in. MultiO is for adults, so before your account opens, please confirm:
By continuing you agree to the Terms of Service and Privacy Policy.
Same conversation — the new model sees everything above.
AI outputs may be inaccurate — verify important information before relying on it.
—
Your payment method and invoices will appear here after your first purchase.
Cancel at the end of the current period? Your plan and access continue until — nothing is cut off today, and you can resume any time before then.
Token rates follow the providers' published list prices, converted at a fixed pivot: 1,000,000 MultiO tokens = $0.15 of standard-tier usage.
| Model | Tier | Input · tokens per 1,000 | Output · tokens per 1,000 |
|---|
Web search: 100,000 tokens per search, shown as its own receipt line.
Add tokens on top of your monthly grant. Purchased tokens are spent after your monthly grant and never expire while your account is active.
Each seat gets its own 80,000,000 tokens every month and access to all models.
Each family seat is $12.00/seat/month and renews automatically until cancelled. Each seat includes its own 80,000,000 MultiO tokens per month at the published token rates; unused monthly tokens expire at the reset and do not roll over. Cancel anytime in Settings → Usage & billing → Cancel subscription; access continues to the end of the paid period.
The seat user needs their own MultiO account first — ask them to sign up, then add them here.
No billing activity yet.
One click sends a prompt into the composer. Templates with {{variables}} ask you to fill them in first. Save your own for the ones you reach for every day.
A persona is a custom assistant: its instructions shape every answer in a conversation that uses it, on any model. Optionally prefer a model and file it under a folder — starting a chat in that folder pre-selects the persona (a default, never a lock). Share a persona with a revocable link; anyone signed in can copy it into their own list.
No personas yet — create your first custom assistant above.
A scheduled task runs its prompt automatically as an ordinary metered chat — or a deep-research run — with the persona you pick, if any. Each run is charged in tokens exactly like sending the prompt yourself, lands in the task's own conversation, and is recorded in the run log below. If your balance cannot cover a run, that run is skipped and logged and the task stays enabled — your balance never goes negative. The shortest schedule is one run per hour; daily times are in UTC.
No scheduled tasks yet — create your first one above.
No runs yet — a run appears here each time a task fires.
Connect a GitHub repository and turn a chat request into a reviewable pull request. The agent reads your code and proposes changes; you approve every file, and you merge on GitHub. It never runs code and never merges.
Loading…
Pick a repository on the left to start a session. The agent works from a read-only snapshot of the branch, and every change it suggests comes back to you as a diff to approve, file by file.
Share your link. When a friend signs up and sends their first few chats, you both get 6,600,000 tokens — automatically. Token credits are spend-only and never expire while your account is active.
Your code: —
Copied.
Facts here are shared with every model — ask MultiO to remember something about you in chat, and it will show up here and be used by GPT, Claude, Gemini, Grok and Llama alike.
Memory is off — these facts are kept but not used or added to.
Facts here are shared with every model — ask MultiO to remember something about you in chat, e.g. “Remember that I prefer short answers.”
Chats in this project use these instructions, the project's files and the project's own memory — never anyone's personal memory. Changes apply to new messages.
Files attached to the project ride along as context in every project chat, for every member — including members who didn't upload them. Your other uploads stay private.
Editors can start chats and attach files; viewers can read the project and its conversations. Everyone pays for their own messages from their own plan — a project shares context, never a balance.
Facts learned from chats in this project, shared by all members. Personal memory is separate and is never used here.
Nothing remembered for this project yet — facts appear here as members chat in the project.
Chats started in this project appear here for every member. Each member's chats stay their own — members can read them in the project, but only the person who started a chat can continue it.
A Teams organisation shares one pooled token balance for its API keys — $25 per seat per month, 80,000,000 tokens per seat into the pool each month. The pool pays only for the organisation’s API keys. Your personal chats and personal API keys always use your own balance, and no admin can ever see them.
Your persistent documents. Write in Markdown, keep every version, rewrite a passage with a model, and export real files — Word, Excel, PDF, Markdown or HTML. Exports are free.
Start a document here, or use “Save as document” on any answer in a chat. Documents are yours alone — they are never shared.
This document contains images — Word, PDF and HTML exports embed them, and Excel places them after the text. An image that can't be included safely (an SVG, or a link that can't be fetched safely) is replaced by its caption text.
Exports are free — generating a file never spends tokens. Word and PDF keep headings, lists, tables and code blocks. Excel turns each Markdown table into its own sheet (Table 1, Table 2, …); a document with no table exports one sheet with one paragraph per row. Images are embedded in Word, PDF and HTML exports and placed after the text in Excel; PDF keeps PNG and JPEG images (GIF and WebP become their caption text there), and any image that can't be included safely is replaced by its caption text.
Every save, AI rewrite and restore adds a version — restoring never rewrites history. The newest 50 versions are kept.
Bring your own key: turns on models from a provider you’ve keyed run on your key and don’t spend your token balance. Web search still costs tokens, and image/video generation still uses MultiO’s keys. Deep-research synthesis follows the same rule as chat; its search steps still cost tokens. Keys are encrypted at rest and are never shown again after saving — only the last 4 characters.
BYOK isn’t configured on this server yet, so keys can’t be saved right now.
Saved.
Connect remote MCP servers (Model Context Protocol) and the models here can use their tools in your conversations. Add a server by its HTTPS URL; if it needs a token, the token is encrypted at rest and is sent only to that server. Read tools (the ones the server marks read-only) run directly. Write tools are blocked unless you turn on write access for that server here.
You are responsible for actions these tools take. Tool results come from the server you connected — MultiO treats them as data, never as instructions, and shows every call in the conversation transcript so you can audit what happened.
MultiO Code connects through the MultiO Code GitHub App — installed per account, on the repositories you choose. MultiO stores no GitHub tokens; disconnecting here uninstalls the App and ends access everywhere. The agent reads a snapshot and proposes changes as pull requests for your approval — it never runs code and never merges.
No activity yet.
Use MultiO from your own code with an OpenAI-compatible API. API usage draws on the same token balance and shows in the same receipt history as chats here. A key is shown exactly once when you create it — after that, only its first characters and last 4 are ever displayed, and a revoked key stops working immediately.
Budget caps are per key and enforced before any provider call — a call that would cross a cap is refused (402 API_KEY_BUDGET_EXCEEDED). Leave a cap empty for no cap. Full documentation: multio.ai/developers.
You won’t see this key again — copy it now and store it somewhere safe.
Copied.
curl https://multio.ai/v1/chat/completions \
-H "Authorization: Bearer $MULTIO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-5.4-nano",
"messages": [{"role": "user", "content": "Hello!"}]
}'List the available models with GET https://multio.ai/v1/models (same Bearer key). Set "stream": true for streaming responses.
Prefer MultiO inside your browser? The MultiO extension adds a side-panel chat plus right-click actions — summarise a page, or ask about selected text — and it signs in with an API key created right here, so it draws on the same balance and shows in the same receipts as everything else. It reads a page only when you click one of its menu items; there is no background page access. Revoke the key here and the extension stops working immediately.
Until the Web Store listing launches, install it the developer way:
chrome://extensions, turn on Developer mode, click Load unpacked and select the unzipped extension folder.A Chrome Web Store listing is planned at launch, which will replace the developer-install steps with a one-click install.
No lock-in: download everything — all conversations, your memory facts, personas, saved prompts, scheduled tasks, share-link records, API-key identifiers (never the secrets themselves), MCP servers you connected (names and URLs only, never the saved credentials) and a billing summary — as one JSON file. Individual conversations can be exported (JSON or Markdown) from the ⋯ menu next to each conversation.
Export downloaded.
Change your password here — you stay signed in on this device, and every other device is signed out. You will get an email confirming the change.
Password changed — other devices have been signed out.
This account uses social sign-in and has no password set. To set one, use “Forgot password” on the sign-in page — the reset link will let you choose a password.
Signed in somewhere you don’t recognise? This signs out every device, including this one, immediately.
Loading…
Add this secret to your authenticator app (1Password, Google Authenticator, Authy…), then enter the 6-digit code it shows to confirm:
Or open this link in the app:
Save these backup codes now — each one signs you in once if you lose your authenticator app, and they are shown only this once:
Two-factor applies to password sign-in. Signing in with a passkey skips this step — a passkey is already two factors (your device plus its unlock). Setting up or removing a credential asks for your current password, so a borrowed session cannot change how you sign in. Accounts that use social sign-in only have no password yet: set one first (sign-in page → “Forgot password”), or rely on a passkey, which needs no password.
Sign in with your device’s fingerprint, face or PIN — no password, no code. Passkeys already on this account:
Delete my account. This cannot be undone. Before you continue, download your data — deletion is permanent:
If you delete your account:
Saved.
AI outputs may be inaccurate — verify important information before relying on it.
That link does not lead anywhere in MultiO.