API Credit Costs

Base render is 1 credit per rendered minute. Transcription adds 1 credit. Caption analysis adds 1 credit.

Face tracking adds 1 credit minimum, or rendered minutes x 1 credit, whichever is higher. Minimum charge is 2 credits.

The same formula applies whether captions are generated from ASR or supplied with SRT.

Every AI Dubbing API request must include durationSeconds. Subclip checks the estimated credit requirement before creating the job.

The preflight formula is transcription minutes when no transcript is supplied, plus dubbing minutes, plus voice clone and background-preservation add-ons when requested.

AI Clipping API jobs add 5 API orchestration credits for source import, API packaging, and final render orchestration.

When you omit segments, the existing AI clipping pipeline costs still apply. When dynamicCaptions is true, Viral Captions analysis costs also apply.

When youtubeTranscript is true and the YouTube transcript is fetched, the completed job adds 1 credit. If transcript fetch fails and the job falls back to ASR, that extra credit is not counted.

Cost is the model's base credits per output x its model-specific output-tier multiplier x output count. The result has a 1-credit minimum and is rounded to 2 decimal places.

Send the exact output tier returned by the selected model in GET /models. GPT Image 2 uses quality tiers (Low, Medium, High); other models may use resolution tiers (1K, 2K, 4K). These values are not aliases and do not imply physical pixel equivalence.

Billing groups are 0.7x for Low; 1x for 1K or Medium; 1.4x for 2K or High; and 2.5x for 4K.

For runtime planning, use the selected model's live estimatedCreditsExamples from GET /models and the estimatedCredits returned when the job starts.

Output count is 1 unless the model supports batch output. Batch models accept 1 to 4 outputs.

Cost is the model's base credits per second x billable duration x the resolution multiplier x the generated-audio multiplier. Video jobs return one output. The result has a 1-credit minimum and is rounded to 2 decimal places.

Resolution multipliers are 1x for 720p, 768p, or source; 1.4x for 1080p or 1440p; and 2.5x for 2160p.

Generated audio adds a 1.2x multiplier when enabled on a supported model. Uploaded audio references do not add this multiplier. Wan 2.7 requires generated audio, so its estimate always includes 1.2x.

Most models use one of their fixed native duration choices below. Gemini Omni Flash is source-video editing only and bills the uploaded source video's matching duration from 3 to 10 seconds. Veo 3.1's three-reference-image mode uses 8 seconds, and Hailuo 2.3 at 1080p uses 6 seconds.

No configured text-to-video or image-to-video model accepts a native 3-second request. For an exact 3.000-second deliverable, trim after download; billing uses the native generated duration before that downstream trim.

For runtime planning, use the selected model's live estimatedCreditsExamples from GET /models and the estimatedCredits returned when the job starts.

Credits are estimated before the job starts. Text to Speech costs 5 credits per generated audio minute with a 1 credit minimum. Voice clone adds 5 credits.

Subclip imports the media first, probes duration, then checks credits before transcription starts.

Cost is 1 credit per source media minute, rounded up to a whole minute.

Without AI analysis or add-ons, cost is 2 credits minimum, or rendered minutes x 1 credit, whichever is higher.

BGM or SFX adds 1 credit total. bgmQuery has no separate search charge.

AI planning adds 2 credits when aiAnalysis is true. Vision analysis adds 0.25 credits per image and 1 credit per video. Audio analysis adds 1 credit per uploaded audio file.

Voiceover adds 1 credit minimum, or rendered minutes x 5 credits, whichever is higher. Stock B-roll search is not exposed in the Video Studio API right now.

API background music separation is disabled. Provider choice does not change the credit formula.

Cost is 1 credit minimum, or duration seconds x 0.10816 credits, whichever is higher. The result is rounded up to the next whole credit.

That works out to about 6.49 credits per minute.