Credits & spend

Every time an agent runs inference it costs money. LEMA Chat meters that usage, prices it, and shows it on the Admin console's Spend page, so a workspace's AI spend is visible and governable.

What's metered

• Usage is recorded at completion of each inference, through a single shared inference path — so live agent turns and every deliberation round are counted the same way.
• Platform-key usage is charged. BYOK usage (a user's or workspace's own key, see Connections) is costed and flagged for visibility but not billed to the workspace — the user is paying their own provider directly.
• The MCP tool surface doesn't run inference, so it isn't credit-priced.

Spend, pricing, and budgets

The console exposes:

• Spend — totals, trend over time, and recent line items.
• Pricing — the per-model prices used to value usage (editable by SuperAdmin).
• Budgets — spend limits.

Editing pricing and budgets, and granting credits, are SuperAdmin actions.

What's not built yet

The metering and reporting are shipped and internal. Not shipped:

• No purchase flow — there's no self-service way to buy credits.
• No credit tiers or expiry buckets — credits aren't packaged into plans.
• The workspace-facing Billing view is a coming-soon placeholder.

Deliberate scope

Credits today are an internal cost-visibility and governance tool, not a commerce system. Buying, tiering, and customer billing are deliberately deferred.