Video API Pricing Compared: Credits vs Per-Minute vs Per-GB

Video processing via API costs anywhere from a fraction of a cent per job to several dollars per minute of output, depending on the provider and pricing model. The wide range makes it hard to compare options without understanding how each model works and where it breaks down. This guide breaks down the three main pricing approaches, compares the major providers, and helps you figure out which one fits your workload.

What are the three pricing models for video APIs?

Video processing APIs use one of three pricing models: credits (per job or per unit), per-minute (based on video duration), and per-GB (based on file size). Some providers blend these — charging per minute for encoding but per GB for storage — but most have a primary model that drives the majority of your bill.

Each model has a different relationship between cost and workload. Credits give you predictable per-job costs. Per-minute pricing scales linearly with video length. Per-GB pricing scales with file size, which is influenced by resolution, bitrate, and codec.

Understanding these differences matters because the cheapest option for a hobby project might be the most expensive option at scale, and vice versa.

How does credit-based pricing work?

In a credit-based model, you purchase credits upfront and spend them when you submit jobs. Each job type has a fixed credit cost (or a cost per minute of output), and you can see exactly what a job will cost before you submit it.

The main advantage is predictability. You know the cost of each operation before it runs. A watermark job costs a known amount per minute of output. A QR code generation costs a flat rate. There is no surprise bill at the end of the month because you saw every charge in advance.

The disadvantage is that credits can feel like an abstraction layer over real money. You have to think in terms of credits rather than dollars, which adds cognitive overhead. Some providers make this worse by using opaque credit systems where the dollar value of a credit is not immediately clear.

Rendobar uses a dollar-denominated credit model, which eliminates the abstraction problem. Credits are denominated in USD, so $1 of credits equals $1 of processing. Job costs range from $0.02 for a QR code to approximately $0.10 per minute for video processing. New accounts receive $5 in signup credits with no credit card required, and Pro subscribers get $5 in monthly credits plus a 20% bonus on every credit purchase.

Creatomate also uses credits, but their system is resolution-based — a 720p render costs fewer credits than a 4K render of the same duration. This makes sense for their rendering-focused use case but adds complexity when comparing across providers.

How does per-minute pricing work?

Per-minute pricing charges based on the duration of the input or output video. A 10-minute video costs 10 times as much as a 1-minute video, regardless of resolution, bitrate, or complexity.

This model is simple to understand and easy to predict if you know your video lengths. It works well for consistent workloads where video duration is the primary variable — transcoding a library of episodes, generating clips from long-form content, or batch processing user uploads of similar length.

The disadvantage is that cost scales linearly with duration even when processing complexity does not. Transcoding a 30-minute podcast at 720p costs 30 times a 1-minute clip, even though the per-frame processing cost is similar. For long-form content, per-minute pricing can get expensive quickly.

Shotstack, Coconut, and AWS MediaConvert all use per-minute models, though their per-minute rates vary significantly (from $0.0075/min for AWS MediaConvert to $0.40/min for Shotstack pay-as-you-go).

How does per-GB pricing work?

Per-GB pricing charges based on the size of the data processed, usually measured as the size of input files, output files, or both. File size is influenced by resolution, bitrate, codec, and duration, making this the least predictable model.

The advantage is that it aligns cost with infrastructure cost. Processing a 4K ProRes file (which might be 50 GB for a short clip) genuinely costs more in compute, storage, and bandwidth than processing a 720p H.264 file (which might be 500 MB for the same duration). Per-GB pricing reflects this.

The disadvantage is unpredictability. Two videos of the same duration can differ in file size by 10x or more depending on codec and bitrate. A user uploading a 4K ProRes master versus an H.264 web export will see dramatically different bills for the same logical operation. This makes it difficult for your product team to quote prices to end users.

Transloadit and Rendi use per-GB models. Transloadit bundles encoding, file handling, and other operations into plans measured by total data processed. Rendi offers a simpler model with monthly GB allowances.

How do the major providers compare?

Here is a side-by-side comparison of the major video processing APIs as of March 2026. Pricing is based on publicly available information from each provider’s website and documentation.

Provider	Model	Starting Price	Free Tier
Rendobar	Credits (per job)	$0.02/job (QR) to ~$0.10/min (video)	$5 signup credits
Shotstack	Per rendered minute	$0.40/min PAYG, $0.20/min subscription	20 min/month
Creatomate	Credits (by resolution)	$41/month (144 min at 720p)	50 credits trial
Coconut	Per output minute	$0.015/min	1 min test videos
Transloadit	Per GB processed	Plans from free to enterprise	Community plan
Rendi	Per GB processed	$25/month (100 GB)	50 GB free
RenderIO	Per command	$9/month (500 commands)	None
Mux	Usage-based	$0.0075/min encoding	$100 credits for $20/mo
Cloudinary	Credits	$89/month (Plus)	Free tier available
AWS MediaConvert	Per minute	$0.0075/min	AWS free tier

Last verified: March 2026. Prices may have changed since publication. Check each provider’s pricing page for current rates.

A few things stand out from this comparison.

AWS MediaConvert and Mux have the lowest per-minute rates at $0.0075/min, but both require significant setup. MediaConvert is a raw AWS service with no high-level API — you write IAM policies, configure S3 buckets, and manage job templates yourself. Mux is primarily a video streaming platform that happens to offer encoding.

Coconut offers very competitive per-minute pricing at $0.015/min, roughly double MediaConvert but with a much simpler API. However, their free tier is extremely limited (1-minute test videos), making it hard to evaluate without committing to a paid plan.

Shotstack’s per-minute rate drops significantly on subscription plans from $0.40/min (pay-as-you-go) to $0.20/min (subscription), but even the subscription rate is among the highest for pure transcoding. Shotstack’s strength is in video composition and template rendering, where the per-minute cost is more justified.

Cloudinary’s pricing is hard to compare directly because their credit system bundles image processing, video processing, storage, and CDN delivery into a single credit pool. The $89/month Plus plan is competitive if you use Cloudinary for your entire media pipeline, but expensive if you only need video processing.

Which pricing model is cheapest for small projects?

For hobby projects, prototypes, and early-stage startups processing fewer than 100 videos per month, the priority is a generous free tier and low minimum commitment.

Rendobar’s $5 signup credit with no expiration and no credit card requirement is the most accessible starting point. At ~$0.10 per minute of video processing, $5 covers approximately 50 minutes of processing. For non-video operations like QR code generation ($0.02 per job), $5 covers 250 jobs.

Coconut at $0.015/min is the cheapest per-minute rate for pure transcoding, but their free tier is extremely limited. If you need to process more than a few test videos, you are paying from day one.

AWS MediaConvert at $0.0075/min is technically the cheapest at scale, but the setup cost in engineering time (IAM, S3, CloudWatch, job templates) means it is not practical for small projects.

Recommendation for small projects: Start with a provider that has a meaningful free tier and no minimum commitment. Rendobar, Transloadit (community plan), and Cloudinary (free tier) all qualify. Avoid providers that require a monthly subscription to get started.

Which model scales best for high volume?

For teams processing thousands of videos per day, the pricing model matters less than the per-unit rate and the overhead of managing the integration.

At high volume, per-minute models (AWS MediaConvert, Coconut, Mux) tend to be cheapest because their per-unit rates are lowest. AWS MediaConvert at $0.0075/min processing 10,000 minutes per month costs $75. The same workload on Shotstack (subscription) would cost $2,000.

However, raw cost per minute is not the only factor. Consider these operational costs:

Infrastructure overhead. AWS MediaConvert requires you to manage S3 buckets, IAM roles, CloudWatch monitoring, error handling, and retry logic. This is engineering time that has a dollar cost. A managed API like Rendobar or Shotstack handles all of this for you.

Concurrency limits. Most providers limit concurrent jobs. If you need to burst to hundreds of concurrent jobs, check whether the provider can handle it without queuing delays. Rendobar’s Pro plan supports 25 concurrent jobs with priority queue access.

Storage costs. Most comparisons focus on processing costs and ignore storage. If your outputs are large and you need to retain them, storage fees can exceed processing fees. Rendobar retains outputs for 7 days (Free) or 30 days (Pro) with no separate storage charges.

Webhook reliability. At high volume, you need reliable job completion notifications. Providers that offer webhooks with retry and delivery tracking (like Rendobar) save you from building polling infrastructure.

Recommendation for high volume: If you have the engineering team to manage AWS infrastructure, MediaConvert offers the lowest per-unit cost. If you want a managed API with predictable pricing, compare Rendobar (credits), Coconut (per-minute), and Transloadit (per-GB) for your specific workload mix.

Which model is best for unpredictable workloads?

Some applications have highly variable workloads — seasonal spikes, viral content, or user-driven uploads where you cannot predict volume or video length in advance.

Credit-based models handle this best because there is no monthly commitment to waste during slow periods. You buy credits when you need them and they do not expire. During spikes, you buy more. During quiet months, your unused credits sit there.

Per-minute subscription models are the worst fit for unpredictable workloads. A $200/month plan that gives you 1,000 minutes is wasted if you only process 100 minutes in a slow month.

Pay-as-you-go per-minute models (Shotstack PAYG, AWS MediaConvert) handle variability well but typically charge a premium compared to subscription rates. This is the cost of flexibility.

Recommendation for variable workloads: Use a credit-based or pay-as-you-go model with no monthly minimum. Rendobar credits never expire and there is no subscription required on the Free plan. Shotstack PAYG and AWS MediaConvert also work, at higher and lower per-unit rates respectively.

What hidden costs should you watch for?

Several costs are frequently omitted from pricing pages:

Egress fees. AWS charges for data transfer out of S3. If you process videos in MediaConvert and serve results to users, egress can add 10-30% to your bill depending on volume and region.

Storage fees. Some providers charge separately for output storage. If your processed videos sit in their storage for weeks or months, this accumulates. Check the retention policy and whether storage is included.

Minimum charges. Some providers round up to a minimum job duration. A 5-second clip might be billed as 1 minute. This matters if you process many short clips (thumbnails, GIFs, previews).

Codec surcharges. AWS MediaConvert charges different rates for different codecs. H.265 (HEVC) encoding costs more than H.264. AV1 costs even more. If your workflow requires newer codecs, factor in the surcharge.

Support tiers. Enterprise support is typically an additional fee across all providers. If you need guaranteed response times, budget for this separately.

How should you evaluate pricing for your use case?

Rather than comparing headline rates, model your actual workload:

Count your job types. How many are simple transcodes versus complex compositions? Simple transcodes favor per-minute models. Mixed workloads with different job types favor per-job credit models.
Measure your video lengths. If most of your videos are under 1 minute (social clips, thumbnails, previews), per-minute pricing is favorable because costs stay low. If you process long-form content (30+ minutes), per-minute costs add up fast.
Estimate your file sizes. If you accept high-bitrate uploads (ProRes, RAW), per-GB pricing will be expensive. If your inputs are already compressed (H.264, WebM), per-GB is more reasonable.
Project your growth. A provider that is cheapest at 100 jobs/month may not be cheapest at 10,000. Model your cost at current volume, 10x, and 100x.
Include engineering time. A $0.0075/min service that requires 2 weeks of setup costs more in the first year than a $0.10/min service you can integrate in an afternoon.

The cheapest provider is the one that minimizes total cost: processing fees plus engineering time plus operational overhead. Sometimes that is the one with the lowest per-unit rate. Sometimes it is the one with the simplest API.