CaptionPipe alternative: caption.sh vs CaptionPipe for API captions

If you are shopping for a captions API you can drive from code, CaptionPipe and caption.sh are the two names that come up with an MCP server attached. They solve the same core problem - send video in, get a captioned MP4 back - but they make different trade-offs on styling, pricing, and file formats. This is the honest version, as of writing, so you can pick the right one for your pipeline.

What CaptionPipe does well

CaptionPipe is a solid, focused product. One call burns captions into your video, and you get the MP4 plus SRT and VTT sidecar files back. If your workflow needs those sidecar files - say you also publish to a player that renders WebVTT - that is a real feature, and it is one caption.sh does not offer. We burn captions into the frames and that is the whole output.

Their pricing is aggressive: $10 buys 250 minutes, so $0.04 per rendered minute, prepaid, and paid minutes never expire. They also offer a custom dictionary (up to 1,000 terms) for getting brand names spelled right, and free re-renders from an edited words.json when the transcript needs a fix. As of writing, their trial is 5 free minutes on videos up to 3 minutes long.

Where caption.sh is different

The short version: styling. CaptionPipe gives you three fixed presets - highlight, clean, boxed - with a hex color for the active word on the highlight preset. caption.sh hands you the whole style layer as JSON: words per group, typeface, size, colors, the active-word animation, position on screen. If your captions need to match a brand kit, a template a customer uploaded, or a look you saw on a viral clip, presets are a ceiling and JSON is not.

That matters most for UGC pipelines and clip tools, where the caption style IS the product surface. Every account you serve wants its own look. With caption.sh the look is data, so your code can generate it per customer, per template, per A/B test, without waiting on a vendor to add a preset.

The agent angle

Both products ship an MCP server, which is why you are probably comparing them at all. caption.sh leans further into agent-native integration: a public OpenAPI spec at api.caption.sh/openapi.json, an llms.txt that maps the whole surface for coding agents, and a listing on the official MCP registry. If an agent is driving the integration, it can read the spec and call the API correctly on the first try.

Pricing, honestly

caption.sh charges $0.10 per rendered minute, credits-based, no tier lock-in. CaptionPipe is $0.04 per minute prepaid. On raw unit price, they win, and we are not going to pretend otherwise. The question is what the extra cents buy: if your captions are a fixed house style, probably nothing, and you should take the cheaper option. If your captions need per-template styling, the comparison is not minute price against minute price - it is minute price against the engineering time of building a styling layer yourself.

When to pick which

• Pick CaptionPipe if you want SRT and VTT sidecar files alongside the MP4, a fixed preset look, and the lowest per-minute price.
• Pick caption.sh if caption styling needs to be fully programmable per template or per customer, or if an agent is the one integrating and you want an OpenAPI spec and llms.txt to point it at.
• Either way, skip the editor-first GUI tools if your videos are produced by software. A UI is a bottleneck when the producer is code.

You can test caption.sh without an account: the playground on the homepage runs a real render in the browser. If the JSON styling does what you need, the API key self-serves from the dashboard.