Start Pilot
[ AI Web Builder Tokens ]

Run open-source AI
only when you need it.
Each request goes to the leanest open-source model that can do the job, runs on the gend.me compute gas network, and costs only the compute it used. No idle servers. No monthly minimum.

Loading your balance…
The leanest open-source model for every request
Runs on the gas network when you call it
Pay only for the compute you use
Buy tokens Manage team balance
0text requests from one $50 pack
0open-source models on the network
—tokens for a typical text request
AI Web Builder Tokens

Buy tokens, spend them request by request

Tokens power every AI feature in your web apps: the chat widget, agents, sequences and the Da Vinci builder. Tokens stay in your balance until your requests use them, and larger packs lower the price per token.

Loading packs…

How a request runs

Spun up for the request. Not kept running.

You never rent a model or a server. Every request takes the same four steps, and you are billed once, at the end, for the compute it actually used.

  1. You make a requestFrom the chat widget, an agent, a sequence or the builder. Nothing is reserved in advance.
  2. The optimizer trims itIt strips everything the request does not need and picks the leanest open-source model that still gives the best result.
  3. The network runs itA connected device serves it if one is available, otherwise gend.me compute does. Either way it runs on the compute gas blockchain network.
  4. You pay the real compute costThe request is billed in tokens at its tracked compute cost. While you are not asking for anything, nothing is spent.
What a request costs

Priced by the compute it takes

These figures come straight from the live token pricing. Small, quick requests cost a fraction of a token. Heavier work such as images and video costs more because it takes real GPU time.

Open-source micro models

Small models, picked for the job

Instead of sending everything to one giant model, the network keeps a catalog of compact open-source models and routes each request to the smallest one that can do it well. Less compute per request means fewer tokens spent.

Team balance

Buy and manage tokens for your team

Pick one of your web apps to see every team member’s token balance. Buy a pack for the people who need one, or send tokens from your own balance.

Sign in to manage your team’s tokens.

Donate your extra compute

Let idle devices earn Gas

The compute that runs AI requests comes from the network. Mobile, desktop and server nodes attached to your web app can serve those requests when they are idle, and earn Gas for it.

Pick a web app aboveto see the mobile, desktop and server nodes attached to it.
Mobile nodesAttach a phoneRun the node app on iOS or Android and earn Gas while it is idle.Get the mobile node → Desktop nodesAttach a computerConnect Mac or Windows machines and let them serve requests off-hours.Download the desktop client → Server nodesAttach a serverConnect racks over the Linux CLI for high-volume network tasks.Read the server docs →
How the gas network works Server upgrades
Ready when you are

Spend tokens on the work, not on idle servers.

Pick a pack, let the network run the right open-source model for each request, and watch the tokens go only where the compute did.

Buy tokens Manage team balance