Lock an AI generation price before the job runs
Pre-generation price locking binds deterministic options, the selected route, quote expiry, credit reservation, a failover ceiling, and post-job reconciliation—before the generation starts.
Locking an AI generation price before the job runs is a pre-execution contract for one image or video request. It is not a subscription plan menu, not a credit-pack top-up, and not a vague “about this many credits” hint. The useful unit is one completed generation with fixed public model identity, fixed material options, a selected eligible route, an expiring signed quote, and a reserved customer charge that cannot rise after approval.
OfflineCreator’s documented router sequence makes that lock explicit. Eligible routes are compared for the exact request envelope, one healthy route is selected, and a signed quote records the selected route, options, rate card, credit charge, and expiry before credits move. Changing a model or material option requires a fresh quote. Studio signs and reserves the customer quote before submission so the approved charge is bound before the provider job starts.
Adjacent industry docs show related but different timing. fal documents that costs can be estimated before a request while also stating that prices vary by model and may change. OpenRouter documents post-request cost calculation from provider token totals with credit deduction after the call. Those patterns explain why a customer-facing lock is a distinct product decision; they do not transfer another product’s prices or settlement rules onto OfflineCreator.
- Pre-execution lock
- Quote before generationApprove a signed charge for one request envelope before the job starts.
- Bound options
- Deterministic settingsDuration, resolution, and other material options are part of the lock, not free variables.
- Not plan shopping
- One job, one reserved chargeSubscription tiers and top-ups answer a different budgeting question.
Failure mode: estimating after spend, or mutating options after approval
Community retrieval for this query was mostly homonyms and macro cost news: NFT “floor price” markets, vendor “lock-in” debates, hiring automation clips, and model sticker-price headlines. The thin usable signal is that practitioners still talk about AI API cost as a routing or timing problem rather than a one-time model pick. Treat those social items as untrusted anecdotes; they do not prove OfflineCreator savings, uptime, or customer outcomes.
The distinctive pre-generation failure mode is a charge that looked fixed and then moved. Options changed after the quote issued. The five-minute signature window expired and reservation still proceeded against a stale envelope. Failover silently selected a more expensive route. Or the product only estimated cost, then settled from post-hoc provider usage the way some external gateways document. fal’s docs allow pre-request estimates while noting prices may change; OpenRouter documents post-request deduction. OfflineCreator’s documented contract is stricter for the customer charge: reserve the signed quote first, keep failover inside the locked envelope, and reconcile actuals without raising the approved price.
Keep the live provider boundary visible. OfflineCreator’s cloud provider disclosure states that the current launch catalog is routed through fal and that supplier admission records the current unit price and billing metric. That supports honest disclosure; it is not a promise that every managed alternate listed in internal routing notes is anonymously live for every account.
- Mutated options
- Old quote is invalidA later duration or resolution change is a new request shape, not a patch.
- Post-hoc settlement contrast
- Estimate ≠ locked reservationExternal docs that debit after usage leave room for a separate customer pre-lock.
- Disclosure check
- Know who runs the jobLaunch-catalog generations currently disclose fal as the serving path.
Deterministic options, selected route, expiry, reservation, failover ceiling, reconciliation
Treat the evidence requirement as six checks that fail independently. Deterministic options ask whether the priced settings are frozen into the quote so a later edit cannot reuse the old charge. Selected route asks whether the signed envelope names the provider path chosen for the fixed public model. Quote expiry asks how long that signature remains acceptable. Credit reservation asks when account credits become bound. Failover ceiling asks whether an alternate route can raise the approved charge. Reconciliation asks what happens when provider-reported actual cost differs from the locked envelope.
OfflineCreator’s hub copy and tracked multi-provider routing notes reviewed August 8, 2026 describe the same chain: sign the selection for five minutes, reserve the signed credit charge, allow at most one safe alternate attempt inside the locked wholesale envelope, then store provider-reported actual cost after completion. Variance above the locked envelope is an operations error absorbed by Studio; the customer charge does not increase after reservation.
That is the decision surface for “lock ai generation price before running.” Cost-aware ranking before the quote issues belongs on AI generation cost routing. Quote payload integrity belongs on signed AI generation quotes. This page owns whether the locked price survives from approval through failover and reconciliation.
- Deterministic options
- Settings are part of the lockA material option change invalidates the prior quote and requires a new one.
- Selected route
- Named at quote timeThe signed quote records the chosen route while the public model ID stays fixed.
- Quote expiry
- Five-minute signed windowReservation should accept only a still-valid signed envelope.
- Credit reservation
- Bind credits before submitStudio signs and reserves the customer quote before provider submission.
- Failover ceiling
- No higher customer chargeAn alternate route must fit the locked envelope after a definitive failure.
- Reconciliation
- Actual cost recorded laterProvider actuals are stored; over-envelope variance is absorbed operationally.
Where to go next
If you need the broad product overview—stable model IDs, Kayak-style booking language, and the full router narrative—use the parent AI Router page. Keep this URL focused on pre-generation price locking for the query “lock ai generation price before running.”
If your next question is how public model identity stays stable while routes change, continue to stable AI model IDs across providers. If you need how wholesale cards become comparable request units before a quote is issued, continue to normalize AI provider rate cards. If you are evaluating image-specific routing surfaces, continue to the AI image generation router. Subscription plans, credit top-ups, and general budget dashboards are out of scope here.
Canonical ownership and evidence boundary
This URL owns pre-generation price locking for “lock ai generation price before running,” including deterministic options, selected route, quote expiry, credit reservation, failover ceiling, and reconciliation. It does not own Studio subscription plans, credit top-ups, or general credit budgeting. The parent /ai-router hub owns the broader AI generation router overview.
Supported product statements are limited to OfflineCreator’s documented router and quote behavior plus the live cloud provider disclosure. Supported external statements are limited to fal’s Model APIs pricing docs and OpenRouter’s FAQ on post-request credit deduction. This draft does not assert that every managed alternate route is anonymously live, does not guarantee wholesale or customer prices from internal modeling notes, and does not invent privacy, retention, benchmark, or outcome claims.