A Reddit cost breakdown claims ComfyUI Cloud's 0.27-credit RTX 6000 figure behaves like per-second billing, implying about $4.62 an hour. Use the math to compare hosted costs against local setups, especially alongside €3,000 workshops and free cinema repos.
The pricing complaint is specific: the Reddit math post says the advertised 0.27-credit RTX 6000 figure behaves like runtime billing, not a flat per-workflow charge. At the same time, another thread surfaced a market for high-ticket ComfyUI training, and the KupkaProd Cinema Pipeline repo gives creators a free local alternative built around the same ecosystem.
The core claim comes from one user's repeated tests on ComfyUI Cloud's default RTX 6000 96GB offering. Their calculation starts at $20 for 4,200 credits, then maps 0.27 credits to roughly $0.00128 per second of GPU time, or about $4.62 an hour.
The important caveat is that this is user-reported behavior, not an official pricing explanation. The same post says ComfyUI Cloud labels the RTX 6000 figure as 0.27 tokens per workflow run, but the author's measurements lined up much more closely with per-second usage.
The cloud math landed next to a different signal from the same community: people are already selling ComfyUI products and workshops to professional teams at €3,000. That thread matters less as outrage than as a snapshot of where the market is going.
In the comments, one practitioner said they built a local ComfyUI asset generator for a company that could produce around 50 game levels with four variants each per hour, while another commenter argued the sale is really expertise, not software. Between those replies and the workshop screenshot, ComfyUI looks a lot less like a hobbyist side tool and a lot more like billable production infrastructure.
The clearest counterpoint is not another pricing debate. It is a repo. KupkaProd Cinema Pipeline is pitched as a free local ComfyUI-based full movie pipeline agent, and the Reddit post claims the nine-minute video in the demo was created in post with fewer than 40 words of prompting.
For AI filmmakers, that is the more interesting split screen: metered hosted GPU time on one side, and increasingly ambitious local ComfyUI automation on the other. The repo itself is still early, but it shows how quickly the ecosystem is packaging repeatable film workflows into something creators can fork instead of rent.
