FFmpeg API pricing, the billing unit decides

Rendobar’s FFmpeg API is the one I would put most workloads on, because per-second credits price short clips and long files by the compute they use, with no subscription, no card and no plan jump. A workload of 400 five-second clips a month costs about $1.10 on Rendobar against $11 to $12 on the entry plans of ffmpeg-api.com and RenderIO and $149 a month on ffmpegapi.net’s monthly Ultra plan. Rendi’s free tier undercuts it at $0 a month, but only if you verify a card and every command finishes inside a minute. Switch to a hundred long files and the runtime caps push Rendi and RenderIO onto plans of $99 to $100 a month, while Rendobar’s $9 Pro plan covers the job length.

I build Rendobar, one of the six, which is why every figure below says where it came from. Competitor prices were read off each vendor’s live pricing page on 2026-09-26. Rendobar’s cost comes from real jobs in the site’s committed job archive, not from the rate card.

Six price lists, five units#

Very Good FFmpeg also runs raw commands, but its served pricing page shows no prices, so it is left out instead of priced at a guess.

The workload, and what one job costs on Rendobar#

The reference workload is 400 jobs a month, each a 5-second 1280x720 H.264 clip of 343.9 KB, the sample file behind most of the site’s FFmpeg measurements. Picking that clip means the Rendobar cost is measured, not modelled. Nearly every FFmpeg-command job in the archive since 2026-08-01 ran on it:

A monthly bill is a sum, so the mean is the figure to multiply: 400 × $0.0027 is about $1.10. The median across every job type on Rendobar is higher, $0.0028, because it includes captions and image generation. It does not apply to this workload. An earlier version of this post priced the workload at about $3 a month from the all-types mean. The FFmpeg-only mean is the right basis, so the figure is now about $1.10.

What 400 short clips cost everywhere else#

Without a card, Rendobar is the cheapest way to keep running this workload. Of the services with a published no-card free tier, it is the only one whose grant still covers it after the first month, and it runs your own command, not a preset. Rendi’s page counts input and output together, and even so this workload uses under 1% of its allowance.

None of the runtime ceilings bind here. The median encode of this clip took under a second on our runners, well inside the 1-minute command cap on Rendi’s free tier and RenderIO Starter. The ceilings start to matter in the next workload.

ffmpegapi.net’s annual Ultra rate is lower than its annual Premium rate for ten times the calls. The page lists them that way on 2026-09-26, so both figures are in the table.

The same six on a hundred long files#

Take the opposite shape: 100 jobs a month, each a 1 GB source with 10 minutes of footage that takes about ten minutes to process.

• Rendi: inputs alone are 100 GB, past the 50 GB free tier. The $25 plan includes 250 GB but caps a command at 10 minutes, no margin for a job of about that length, so the safe choice is an unlimited-runtime plan from $100 a month.
• RenderIO: 100 commands fit any plan, but Starter caps a command at 1 minute and Growth ($39) at 5. Ten-minute jobs need Business at $99, where the cap is 90 minutes.
• FFmpeg Micro: 1,000 minutes of source is 6,000 tokens. That fits Pro at $49.99 (7,000 tokens, 2,048 MB input cap) and not Starter.
• ffmpeg-api.com: GB-seconds multiply file size by processing time. One job is at least 1 GB × 600 s, so 100 jobs are over 60,000 GB-seconds, twelve times the 5,000 in its $99 Growth plan.
• ffmpegapi.net: 100 calls fit Premium at $49, if its task endpoints (merge, concatenate, picture-in-picture) cover the work.
• Rendobar: the free plan caps input files at 500 MB, so a 1 GB source needs Pro ($9/mo, 10 GB uploads with 8 GB processed per job, 9-hour jobs). Of the plans that publish a runtime limit, it is the cheapest that accepts a ten-minute job, by about $90 a month. I have not measured a job this size, so there is no compute figure here. Multiplying a 5-second clip up to ten minutes would be the guesswork this post argues against.

The unit that was trivial on short clips, GB-seconds, is now the one that runs off the published price list. The unit that cost the most on short clips, per call, is now among the cheapest.

What each unit meters#

Per-command plans are the easiest to budget, and they are why the runtime cap matters so much: when every command costs the same, the plan limits how long a command may run instead.

Subscription or credits#

Five of the six need a subscription for anything past the free tier. The quota resets monthly, and a quiet month still costs the full plan.

Rendobar sells prepaid credits, and none of them expire, the signup grant, purchased top-ups and Pro’s monthly credits alike. The $5 signup grant needs no card. The $9/mo Pro plan adds $5 of monthly credits and a 25% bonus on purchased credits, but mostly it buys limits: 20 concurrent jobs instead of 1, a 9-hour job timeout instead of 1 hour, and 10 GB input files instead of 500 MB. The $29/mo Studio plan adds $20 of monthly credits with the same 25% bonus. (Corrected on 2026-10-20: this said Pro’s bonus was 20%. It became 25% when the Studio plan launched.) A quiet month costs nothing and a busy one draws down what you already bought, so bursty or seasonal work pays less on credits. The same balance also pays for captions, timelines, compression and image jobs, so a product that grows past transcoding does not add a second bill.

Nothing here measures reliability, speed or support. Those differences are in the best FFmpeg API guide, and what each service runs before you pay is in free FFmpeg API tiers.

Put your real file sizes and runtimes through the first table, and check the runtime cap before the quota. For most teams that points to Rendobar, where the bill follows the compute, no card is needed to start, and a long job never forces a plan change you did not plan for.