Does the Sume run spend cap include the LLM turn? No, only generation
On a Sume run receipt, billable_amount_usd_micros is generation spend only; the agent's LLM turn bills a separate Agent wallet. How to budget both.
If you set generation_spend_cap_usd to 5 on an Agent Completion, you have capped what the run can spend on generation: clips, images, voice. You have not capped the tokens the agent itself uses to think and call tools. Sume's Runs and results page says so plainly for run receipts.
What the receipt says
The usage object on a run holds currency, billable_amount_usd_micros and generation_spend_cap_usd_micros. The docs define the billable amount as the generation spend of the run, the same total Sume uses to enforce the cap, including reserved and captured amounts. It rises while the run is in flight and settles at the end. Two limits are stated: it excludes the agent's LLM turn, which bills the separate Agent wallet, and it is a receipt figure, not an invoice. GET /v1/usage remains the authoritative billing record.
usage can also be null, meaning Sume could not read the spend. That is different from 0, which means the run truly spent nothing yet.
| Question | Answer | Where |
|---|---|---|
| Does the cap bound LLM tokens? | No | Runs and results, usage |
| Is billable_amount an invoice? | No, use GET /v1/usage | Runs and results, usage |
| What does null usage mean? | Spend could not be read | Runs and results, usage |
| Is a cap required on Agent Completions? | Yes, no default | Agent Completions |
How to budget with that
Plan two numbers per run. The first is the cap, which is a hard ceiling on generation. The second is an estimate for the agent turn, which you reconcile later from your usage records. For a nightly batch, multiply both by the number of runs and compare against the wallet before the first one starts.
For a schedule, the default cap is $1.00 when none is set, and a caller can lower the cap for one run but never raise it. Read Create a schedule for the exact rules, and the Agent Completions page for the required field on the ad-hoc path.
Sources
Related posts
More in Agents
- Fire a Sume schedule from a decision, not a clock: API trigger
Set api_trigger_enabled on a Sume schedule, let Clef or Decider decide when it is worth running, then POST /v1/actions/{action_id}/runs with an idempotency key.
- Five generate_image calls in one turn: Sume MCP create budget
Hosted Sume MCP gives paid creates their own budget (20 per principal, 64 per process) and queues a call up to 20 seconds before wait_busy.
- usage.cap on a Format run: an agent reads remaining_usd_micros
A Format run receipt splits its spend cap into limit, counted and remaining USD micros. A supervising agent can read headroom before it asks for more work.
- Sume hosted MCP is not Studio Agent: the one URL to put in your client
Put https://mcp.sume.com/mcp into Claude, Cursor or Codex. Studio Agent is the in-app product, not a customer MCP server, so there is no second URL to add.
Written by Sume