Runway API Pricing Explained
Quick Answer
Runway API usage is billed in credits. The current developer pricing page values one credit at $0.01, so video cost equals credits per second multiplied by duration and then by $0.01. Model rates differ substantially: Gen-4.5 currently costs 12 credits per second, while Gen-4 Turbo costs 5 credits per second. API credits are separate from credits purchased in the Runway web application.
Runway API Credits vs USD
The Runway Developer API uses credits as its billing unit. The pricing page as of June 15, 2026 values one credit at $0.01. This means:
1 credit = $0.01
1,000 credits = $10.00 Video cost formula:
video_cost_usd =
credits_per_second
× generated_duration_seconds
× $0.01 For models with a minimum credit charge per generation:
billable_credits =
max(model_minimum_credits, credits_per_second × duration) Sales tax may be added separately. Rates change; check the official Runway pricing page before production use. Do not substitute Web App subscription credits for Developer API credits.
Current Video Model Rates
The following rates were reviewed on June 15, 2026 from the official Runway Developer API pricing page. Rates vary significantly by model. Always verify the current rate before estimating a project budget.
| Model | Credits per second | Notes |
|---|---|---|
gen4.5 | 12 | Current primary model |
gen4_turbo | 5 | Lower cost alternative |
aleph2 | 28 | 56-credit minimum per generation |
act_two | 5 | Character animation |
seedance2 480p/720p | 36 | Standard quality |
seedance2 1080p | 40 | High quality |
seedance2_fast | 29 | Fast mode, 480p/720p |
veo3 | 40 | |
veo3.1 with audio | 40 | |
veo3.1 without audio | 20 | |
veo3.1_fast with audio | 15 | |
veo3.1_fast without audio | 10 |
Some models listed above are available through the Runway API but are not Runway-native
models — check the official pricing table for the current model roster and which are
Runway-native versus third-party integrations. Note that certain models such as
gen4_aleph and gen3a_turbo are marked deprecated and
scheduled for sunset on 2026-07-30; do not build new integrations around them.
Calculable Cost Examples
Gen-4.5 (12 credits/second)
- 5 seconds: 12 × 5 = 60 credits = $0.60
- 10 seconds: 12 × 10 = 120 credits = $1.20
Gen-4 Turbo (5 credits/second)
- 5 seconds: 5 × 5 = 25 credits = $0.25
- 10 seconds: 5 × 10 = 50 credits = $0.50
Aleph 2.0 (28 credits/second, 56-credit minimum)
- 2-second calculated cost: 28 × 2 = 56 credits = $0.56 (equals minimum)
- 1-second calculated cost: 28 × 1 = 28 credits → billable = 56 credits minimum
These examples apply the listed credits-per-second rate to the generated duration. If a workflow invokes separate image, audio, upscale or additional generation operations, calculate those operations separately using their own official rates.
Web App Credits vs Developer API Credits
Credits or subscription plans purchased at app.runwayml.com are separate from credits purchased at the Developer Portal (dev.runwayml.com). This has practical consequences:
- Web App credits do not appear in your Developer API organization
- Developer API credits do not appear in your Runway Web App workspace
- Confirm you are viewing the Developer Portal organization when checking API credit balance
- Do not purchase a Runway web subscription plan to fund API-based integrations
Minimum Purchase and Autobilling
Runway Developer API billing includes the following terms as of June 15, 2026:
- Initial credit purchase minimum: $10 (1,000 credits)
- Autobilling recharge minimum: 1,000 credits / $10
- The system checks the credit balance periodically — not before every task
- Payment failures follow a retry process documented in the Developer Portal
- Your usage tier's monthly-spend allowance may limit the autobilling recharge amount
Because the autobilling balance check runs periodically rather than before every task, applications should monitor remaining credits and stop or queue work when the balance is low. Do not assume autobilling is instantaneous or certain to recharge before a balance reaches zero.
Usage Tiers and Concurrency Limits
Runway API usage tiers control multiple dimensions:
- Maximum concurrency
- Maximum generations in a rolling 24-hour window
- Maximum credit purchases in a rolling 30-day window
- Per-model or per-modality restrictions
When the concurrency limit is reached, tasks may be marked THROTTLED. A
throttled task has been saved but has not yet entered the processing queue. Do not
immediately create a duplicate submission upon seeing THROTTLED; this
compounds the queue and may cause duplicate outputs.
Runway does not publish a universal maximum requests-per-minute cap. However, tier limits on concurrency, rolling 24-hour generations and rolling 30-day credit purchases still apply. Do not characterize the API as having unrestricted parallel task capacity or no workload constraints.
Task Failures and Credit Treatment
Runway API tasks report a failureCode when a task does not complete
successfully. Credit treatment varies by failure type:
| failureCode | Credit behavior | Recommended action |
|---|---|---|
SAFETY.INPUT.* | Credits are not refunded | Do not retry |
SAFETY.OUTPUT.* | Credits are refunded | Review output failure; retry only when appropriate |
INTERNAL.BAD_OUTPUT.* | Credits are refunded | Correct prompt or input where possible, then retry |
INPUT_PREPROCESSING.SAFETY.TEXT | Credits are refunded | Do not retry unchanged |
INPUT_PREPROCESSING.INTERNAL | Credits are refunded | Wait, then retry |
ASSET.INVALID | Credits are refunded | Correct the input; do not retry unchanged |
THIRD_PARTY.UNAVAILABLE | Credits are refunded | Wait before retrying |
INTERNAL or null | Credits are refunded | Wait, then retry |
Runway's current task-failure documentation says SAFETY.INPUT.* failures
are not refunded, unlike other task failures. Always inspect the returned
failureCode before deciding whether to retry.
What to Log
For every Runway API task, record:
- Task ID
- Status (
succeeded,failed,THROTTLED, etc.) failureCodeand failure message, if applicable- Model used
- Requested duration
- Submitted timestamp (UTC)
- Completion or failure timestamp (UTC)
- Estimated credits before submission
- Actual credit usage from the organization record
- Retry count for this task
Task Lifecycle and Duplicate Cost Risk
Runway API uses asynchronous tasks. Common patterns that lead to unintended cost:
- Duplicate submission — submitting the same task more than once before checking status
- Automatic SDK retry overlapping with application retry — both fire, doubling the charge
- Worker duplicate consumption — a worker picks up the same task twice
- Timeout-triggered new task creation — client times out before a slow generation completes, then creates a new task instead of polling the original
- Multi-output workflows — generating extra outputs without tracking them
- Completion recorded locally but not persisted — a successful generation is re-requested because the result was lost
Recommended workflow:
- Create the task and persist the returned task ID
- Poll or use the SDK
waithelper to monitor completion - Inspect the
statusfield before taking further action - If a timeout occurs, query the original task ID before creating a new one
- Retry only when the
failureCodepermits it
Worked Project Estimate
Planning 200 Gen-4.5 videos at 5 seconds each
Credits per video: 12 × 5 = 60 credits
Total credits: 200 × 60 = 12,000 credits
Estimated API cost: 12,000 × $0.01 = $120 Usable-output cost ratio
cost_per_usable_video =
total_consumed_credits × $0.01
÷ usable_completed_videos Do not automatically include all failed tasks in or exclude them from the denominator. Use the actual credit usage reported in your Developer Portal organization record, and divide by the number of videos that meet your quality and completion criteria.
Image and Other Runway API Models
The Runway Developer API also covers non-video capabilities. These are billed in credits with different units:
gen4_image: 5 credits (720p), 8 credits (1080p)gen4_image_turbo: 2 credits (any resolution)gpt_image_2: 1–41 credits depending on quality, resolution and output count- Upscale, audio and realtime models: varies by capability and resolution
For image-specific cost guidance, see the Image Generation API Cost and GPT Image API Cost guides.
Cost Control Checklist
- Run a representative small test before committing a full project budget
- Fix the model and duration for consistent cost estimation across a batch
- Persist task IDs immediately after submission
- Implement idempotent submission — check for existing task before creating a new one
- Set a maximum retry count per task to prevent runaway charges
- Retry only when the
failureCodeallows it; never retrySAFETY.INPUT.* - Monitor credit balance and pause work when it falls below a safe threshold
- Set the autobilling threshold high enough to cover peak workload between checks
- Track tier limits for concurrency, rolling 24-hour generation windows and rolling 30-day credit purchase caps
- Separate completed, failed and throttled task records to analyze real usable-output cost
- Use actual credit usage from the Developer Portal rather than local estimates for financial reporting
- Note the pricing review date and re-check before any production scaling
Official Sources Reviewed
The information on this page is based on Runway's official developer documentation reviewed as of June 15, 2026. Pricing, model availability and API terms may change; always verify current rates in the Developer Portal before scaling.
- Runway Developer Docs: Pricing
- Runway Developer Docs: Setup and Authentication
- Runway Developer Docs: Autobilling
- Runway Developer Docs: Usage Tiers
- Runway Developer Docs: Task Failures and Error Handling
- Runway Developer Docs: Using the API
- Runway Help: Runway API FAQs
Related Guides
Video Generation API Pricing
Learn more about this topic
Small Prepaid Test
Learn more about this topic
API Billing Mismatch
Learn more about this topic
Failed Generation Cost
Learn more about this topic
Billing Transparency
Learn more about this topic
Image Generation API Cost
Learn more about this topic
GPT Image API Cost
Learn more about this topic
Runway API is billed in credits at $0.01 each, with video cost determined by model, credits per second and generated duration. Gen-4.5 costs 12 credits/second ($0.12/second) while Gen-4 Turbo costs 5 credits/second ($0.05/second). Aleph 2.0 applies a 56-credit minimum. Credits purchased at the Web App are separate from Developer API credits. Autobilling checks balance periodically, not in real time. Task failure credit treatment depends on failureCode — SAFETY.INPUT.* failures are not refunded. Use the Developer Portal organization records for financial reconciliation, not local task estimates. AICostPlanner is an independent educational cost-planning site and is not affiliated with Runway.
Frequently Asked Questions
How much is one Runway API credit?
One Runway API credit equals $0.01 as of June 15, 2026. This rate may change. Multiply the credits consumed by $0.01 to get the USD cost.
How much does a five-second Gen-4.5 video cost?
Gen-4.5 is billed at 12 credits per generated second. A 5-second video consumes 12 × 5 = 60 credits, which equals 60 × $0.01 = $0.60. This estimate applies the official Gen-4.5 rate of 12 credits per generated second. Separate Runway capabilities or additional generation tasks should be estimated independently.
Is Gen-4 Turbo cheaper than Gen-4.5?
Yes. Gen-4 Turbo is billed at 5 credits per second versus Gen-4.5 at 12 credits per second. At 5 seconds, Gen-4 Turbo costs 25 credits ($0.25) compared to Gen-4.5 at 60 credits ($0.60). Actual cost savings depend on whether Gen-4 Turbo meets the quality and capability requirements of your use case.
Are Runway Web App credits usable in the API?
No. Credits and subscription plans purchased at app.runwayml.com are separate from Developer API credits at dev.runwayml.com. API credits do not appear in the Web App workspace and vice versa. Always confirm you are in the correct Developer Portal organization when checking your API credit balance.
Does Runway refund credits for failed tasks?
Runway's current task-failure documentation says SAFETY.INPUT.* failures are not refunded, unlike other task failures. Always inspect the returned failureCode before deciding whether to retry.
What does THROTTLED mean?
THROTTLED means the task was accepted and saved, but has not yet entered the processing queue because the concurrency limit for the current usage tier has been reached. The task will be processed when capacity becomes available. Do not create a duplicate submission upon seeing THROTTLED — wait for the original task to be picked up, or check its status before creating a replacement.
How does Runway autobilling work?
Autobilling automatically recharges your credit balance when it falls below a configured threshold. The minimum recharge amount is 1,000 credits ($10). The balance check runs periodically — approximately hourly — not before every task. Applications should still track their own credit balance and pause or queue work when it is low, rather than relying on autobilling to trigger before a balance runs out.
Is Runway API pricing the same as a Runway subscription?
No. The Runway Web App subscription plans and credit bundles are separate from Developer API pricing. API rates, models and billing are managed in the Developer Portal. Check the official API pricing page for the current credit-to-USD rate and per-model credit costs.
Estimate before you scale
Use a small prepaid test to measure real Runway API cost before committing a larger project budget.