For coding agents and pipelines

The model account for coding agents and pipelines.

Code, evals and production jobs, on one key: Claude, GPT, Gemini and DeepSeek, pay as you go, with a budget and a usage record per key. Connect Claude Code, Codex or any SDK.

Sign up freeSee the console10% extra on your first top-up, up to $10
/console/project/membersdemo data
Workspace
Overview
Models
API keys
Usage and billing
Organization
Members and quota
Settings

Members and quota

Acme Inc · August$0.00Account balance　$760.00 spent this month
MemberSpentQuota
JJessyAdmin$96.00No capChange
WWenMember$91.0091% used · $100 capChange
KKaiMember$12.0024% used · $50 capChange
Nnightly-batchService account$15.0050% used · $30 capChange
loadingmodels available
loadingupstream providers
loadinghealthy right now
What it solves

One account, a key for every agent

Once a team runs more people and more agents, someone has to keep track of the models, keys, budgets and bill.

01

No more signing up with every provider

Claude, GPT, Gemini and DeepSeek all live in lazu. Fund it once and the whole company can use any of them — no separate accounts, top-ups or invoices.

  • New models appear in the list on release day
  • Prices are published; you pay for what you use
Models
Loading the catalogue…
02

Give everyone a cap that actually stops

$50 for a new hire, $30 for a script. When the cap is reached the requests stop — you do not find out from next month's invoice.

  • You get a heads-up before someone hits the cap
  • Raise or lower it any time; it applies to the next request
Change quotaWen
Cap for this month
$30$50$100$200No cap
Spent so far
$91.00$9.00 left — close to the cap
03

One key per person, so a leak costs one key

No shared company key. If someone's key ends up in a public repo, disable that one — everyone else keeps working.

  • Each key can be limited to specific models
  • Revoke everything a leaver held in one action
API keys7 active
Wensk-lazu-4b7c…Active
CIsk-lazu-9d21…Active
Internsk-lazu-1f08…Disabled
04

A bill you can read and hand to finance

Who, which model, when, how much — line by line. Group by person or model and export it as a spreadsheet.

  • Every call recorded, nothing sampled
  • Group by team or project
Usage and billingAugust · by member
design-team$261.00
Jessy$96.00
Wen$91.00
ci-runner$41.00
Getting started

Three steps, running in ten minutes

1Sign up and fund

Signing up creates a team account. Top up from $10; your first top-up gets 10% extra.

2Configure your gateway

Choose the models and routes you want to use in the console.

3Create a key

Paste the key into your tool and see its usage in the console.

FAQ

Before you start

Do I need a card?

Not to sign up. You can look around and create Keys first; before making calls you top up, from $10, and your first top-up gets 10% extra (up to $10). Top-ups aren't refundable.

What happens when a cap is reached?

That person's requests are rejected with a clear “quota exhausted”, and nothing more is charged. Raise the cap and the next request goes through. Nobody else is affected.

How are model prices set?

Model usage is billed at the published price for the selected stable or agent lane. The console shows the applicable lane price before you send traffic.

We already use another provider — is moving painful?

You do not have to move all at once. Send a small slice of traffic through lazu, compare cost and latency, then shift more if it holds up.

Hand out this month's quota

Two minutes to sign up, 10% extra on your first top-up.

Sign up freeRead the docsStar on GitHub