Seat licenses
Pay for access
Every employee is a monthly line item whether they ask one question or none. New models mean new vendors, new keys, and another invoice.
All models. One route.
Company chat and APIs through one mesh: usage billing, 80 models on a single invoice, security governance, and observability your finance team can actually read.
Register with work Microsoft 365, Google, or email. New companies get a $10 promo. Existing companies sign in the same way.
Billing
You pay for tokens the company actually runs. Adding a person does not add a license. Idle people cost nothing.
Control
Work SSO, tenant isolation, secret and PII guardrails, and US or EU residency — before a prompt ever hits a model.
Catalog
OpenAI, Anthropic, Google, DeepSeek, and more through one workspace and one invoice. No frontier API keys on laptops.
Visibility
See who used which model, the spend mix, idle vs heavy people, and daily trend — so finance and IT are not guessing.
Pricing
ChatGPT-style seat plans charge every name on the roster. MeshPool rates input and output tokens. A person who does not use AI does not show up on the bill.
Seat licenses
Every employee is a monthly line item whether they ask one question or none. New models mean new vendors, new keys, and another invoice.
MeshPool
One wallet, one invoice, every model. Optional spend caps are governance — they stop runaway use; they are not the price.
Every alias on the mesh has a published input and output price (image models meter per image or image tokens). The full card is below.
Chat and API traffic hit the same catalog. Finance sees one statement instead of OpenAI plus Anthropic plus whoever else the team signed up for.
Default workspaces top up a wallet. Larger teams can settle on invoice — same rating, different collection. Get a demo if you need packaging.
Rate card
Fast / Balanced / Max are the chat defaults. Named aliases are the same catalog and rates as Model policy — default self-serve markup, what the wallet is billed, in USD per million tokens unless noted.
Default chat
$0.0592 in$0.2720 out
meshpool-fast
Everyday work
$1.20 in$3.84 out
meshpool-balanced
Hard reasoning
$2.08 in$4.16 out
meshpool-max
USD at the default self-serve rate, catalog 2026-09-05.2. Token prices are per million. Image models bill per image or per image-output tokens as shown. Company contracts can differ. Not a quote from the model labs.
Security & governance
Identity comes from your directory. Each company is its own workspace. Admins decide what is allowed to leave chat, and where inference is allowed to run.
Microsoft 365, Google, or email — one issuer per company. Membership, spend, and chat stay scoped to that tenant. Cross-tenant access is refused.
Security & Governance blocks common secrets (API keys, bearer tokens) and PII patterns (email, phone, cards, and more). Admins review filters and toggle rules per workspace.
Choose GLOBAL, US, or EU for where inference may run. EU workspaces pin traffic to residency-safe routing so completions follow the tier you set.
Observability
Most company chat is a black box: a seat count and a hope. MeshPool meters every request, so you can answer who used what, what it cost, and who is not using it at all.
Totals, tokens, and requests with a daily or weekly trend. Break spend down by model so you can see whether Fast is doing the work or Max is running away.
Per-person utilization, idle vs heavy, and adoption in the last seven days. Intensity ranking is usage against peers in your company — not an HR score.
Finance gets a statement from one system instead of scraping four vendor dashboards. Admins can export usage for the window they care about.
Contact
Packaging, residency, or a rollout to the company — email us. Prefer a walkthrough? Use Get a demo in the header.
Register
First login with a company work email opens your MeshPool workspace. Personal and disposable inboxes are not accepted. If the domain is already claimed, we notify your admin. Microsoft 365, Google, or email — same buttons either way.