Skip to main content
Docs

Search guides and API endpoints, for example “Idempotency-Key” or “submit job”.

    Concepts

    Plans and limits

    What each plan includes, and what happens when you reach a limit.

    Updated October 1, 2026

    Every account is on one plan. Paid plans are billed per seat each month and include wallet credit; compute is still charged to the wallet at the plan's rates, which fall as plans go up. Compare plans on the pricing page, and subscribe or change plans on your billing page.

    Plan limits
    LimitBasicStarterProEnterprise
    PriceStarting at $0$100 / month per seat$500 / month per seatContact us for pricing
    Wallet credit each monthNone$50$250None
    Jobs per monthUnlimitedUnlimitedUnlimitedUnlimited
    Jobs running at once (CPU / GPU)1 / 15 / 150 / 5500 / 50
    Utility calls per month50050,000500,000Unlimited
    Inference molecules per month5010,000100,0001,000,000
    API keys151550
    Custom models010100500
    Artifact storage1 GiB5 GiB25 GiB100 GiB
    Artifacts kept for7 days30 days90 days90 days

    When you reach a limit

    LimitWhat happens
    Jobs running at onceExtra jobs wait in queued and start as others finish. Nothing fails.
    Utility calls, inference moleculesSubmits return 403 until the next calendar month. Get usage and limits shows this month's counts.
    API keysCreating another key returns 403 (api-key-limit-exceeded). Keys you already have keep working after a downgrade.
    Custom modelsTraining another model fails until you delete one.
    Artifact storageUploads that don't fit are refused, and a workflow run that needs more space pauses until you free some and resume it. Workflow estimates warn when you pass 80% of your quota. See Artifacts.
    Artifact retentionResults older than your plan's retention period are deleted. Download anything you want to keep.

    Jobs are not capped per month: your wallet balance and the concurrency limit are what bound them.

    Request rate limits are the same on every plan; see Rate limits.