LUNAROUTETEAMS[ GET API KEY ]
// LUNAROUTE TEAMS

Plans for teams that
ship with agents.

Per seat, priced by the lane. Every seat ships with concurrency lanes plus background lanes for batch work — and the enterprise admin to govern who gets what. Add lanes à la carte as you grow; over your allowance you're never blocked, just gracefully degraded.

SSO, audit logs, per-user & per-key lane assignment
virtual keys + budgets, routing for OpenAI/Anthropic/etc.
NO LOGS, every model, one endpoint
MARE SERENITATIS
PER SEAT · PRICED BY THE LANE
OCEANUS PROCELLARUM
SSO · AUDIT · PER-KEY LANES
MARE NUBIUM
EVERY MODEL · ONE ENDPOINT · NO LOGS
// 01 - PER SEAT

Per seat, priced by the lane.

One simple seat price covers the platform and a baseline of compute. Every seat includes 2 concurrent lanes + 2 background lanes per seat — enough for an engineer and an agent working in parallel, with a background pool for batch jobs. Need more? Add lanes à la carte.

TEAMS · PER SEAT
$39/seat/mo
billed monthly or annually · cancel anytime · no long-term contract
2 concurrent lanes + 2 background lanes per seat
SSO (SAML/OIDC) + SCIM provisioning
Audit logs (API + export)
Per-user and per-API-key lane assignment
Virtual keys + per-key budgets
Routing for OpenAI, Anthropic & other providers
NO LOGS · every model · one endpoint
GET API KEY →

A lane is room for one inference request in flight. Concurrent lanes cap interactive parallelism; background lanes are a separate pool for batch and ballast work that can wait. Over your lane-hour allowance you're never blocked — traffic degrades to a lower QoS tier and keeps running.

teams — seat assignment
$ lunaroute-cli teams seats assign \
    --user ada@yourco.com \
    --lanes 4 --bg-lanes 4 \
    --budget 200   # $/month cap

$ lunaroute-cli keys create ci-nightly \
    --lanes 2 --bg-lanes 8 \
    --budget 80

✓ assigned. concurrency and background
lanes are enforced per user and per key.
// 02 - À LA CARTE

À la carte — add lanes as you grow.

Start every seat at the baseline and dial up the teams, agents, or keys that need more. No tier lock-in; mix and match across seats and keys.

More concurrent lanes
+$10/seat/mo per lane
Add interactive parallelism where it matters — power users, busy agents, CI fan-out. Assigned per seat or per API key.
More background lanes
+$5/seat/mo per lane
Cheaper than interactive lanes. A separate pool for batch eval, indexing, and ballast work that can wait for quiet GPU capacity.
Keep-serving-priority lane-hours
$1.75/lane-hour
Rather degrade later than now? Add lane-hours at $1.75/lane-hour to keep serving priority past your allowance — same rate as our Ultra plan.
// 03 - ADMIN

Enterprise admin, on every seat.

The controls a platform team needs to hand AI to a whole engineering org — included from the first seat, not gated behind a "contact sales" enterprise tier.

SSO (SAML/OIDC) + SCIM provisioning
Federate identity from your IdP; users appear and disappear with your directory, not a spreadsheet.
Audit logs
Every admin action, key rotation, lane/budget change, and request recorded. API + export for your SIEM.
Per-user and per-API-key concurrency + background lanes
The control OpenAI and Anthropic don't offer: assign exact lane counts per user and per key, not just a spend cap.
Virtual keys + per-key budgets
One key per teammate, agent, or CI job — each with its own lanes, background lanes, and a hard monthly budget. Revoke in one click.
Routing for other providers
Route to OpenAI, Anthropic, Gemini, and the rest behind one OpenAI-compatible endpoint. BYO keys or use ours; failover without touching agent config.
NO LOGS, at team scale
Request and response bodies processed in memory, never persisted — not by us, not by our providers (it's in the contracts). Your ledger is the whole record.

Bring your team aboard.

$39/seat/mo, 2 + 2 lanes to start, and the admin to govern it. Add lanes as you grow; over-allowance never blocks.