RunningHub
Run ComfyUI image and video workflows in the cloud without installing a thing.
Overview
RunningHub bills itself as an all-in-one AI content creation platform powered by native AI agents, which is a lot of words for a fairly practical idea: take ComfyUI, the node-based workflow tool that serious generators swear by and casual users bounce off, and run it on someone else's hardware. You open a workflow, hit run, and it executes in the cloud. Around that sit AI Apps for one-click jobs like upscaling, image-to-anime, text-to-video and image-to-video, a Quick Create mode, an Explore gallery of other people's work, and a developer API aggregating over 400 models behind one interface.
Video is the larger half of the catalogue, with roughly 322 video models against 189 image models, and recent releases land quickly: Nano Banana 2.1, Seedance 2.5, Vidu Q4 Preview and MiniMax H3 are all live. The technology is genuinely capable. The friction is the money. Platform compute is billed per second in a virtual currency called RH Coins, and RunningHub does not publish what an RH Coin costs, nor what a membership costs, to anyone who is not logged in. The developer API is priced openly in plain dollars, which makes the opacity on the subscription side harder to excuse rather than easier.
Key Features
Cloud ComfyUI workflows: Run node-based ComfyUI workflows in one click, with no local install, GPU or dependency setup.
Video-first model catalogue: Around 322 video models alongside 189 image models, covering text-to-video, image-to-video and recent releases such as Seedance 2.5, Vidu Q4 Preview and MiniMax H3.
AI Apps: One-click tools for image generation and enhancement, upscaling, image-to-anime, text-to-video and image-to-video.
rhTV infinite canvas: A freeform canvas for assembling and iterating on generations.
Model API: A single standard interface to 400+ models across video, image, audio and 3D, plus LLM, AI application and workflow endpoints.
Quick Create and Explore: A fast-path creation mode and a public gallery of community workflows and outputs.
Open-source ComfyUI plugins: Official plugins on GitHub for calling RunningHub workflows from a local ComfyUI install.
Creator rewards and referrals: An invite programme granting RH Coins to both the inviter and the new user.
Pricing
Starting price
Usage-based (plan prices not public)
Checked 2 November 2026. RunningHub publishes some prices openly and keeps others behind a login, so read this section carefully before committing money.
Model API, published openly: Priced in US dollars at runninghub.ai/call-api, per second, per call, per image or per million tokens depending on the model. Observed list prices that day ran from roughly $0.01 per call up to $30 per million tokens, most shown alongside a slightly discounted rate. No account is needed to see these.
Promotional per-second rates on the homepage: Nano Banana 2.1 from $0.008/sec, MiniMax H3 at 480P from $0.015/sec (50% off for members, 27 October to 26 November), Vidu Q4 Preview from $0.023/sec and Seedance 2.5 from $0.036/sec. These are time-limited promotions, not standing rates.
Platform compute is billed per second in RH Coins. RunningHub's own app pages state the Standard tier (24GB VRAM) consumes 0.2 RH Coins per second and the Plus tier (48GB VRAM) consumes 0.4 RH Coins per second.
Free daily allowance: Multiple independent sources and community posts on RunningHub's own site report 100 RH Coins credited on each daily sign-in, valid 24 hours, one task at a time, with a 20-minute cap. At the Standard rate that is roughly eight minutes of compute a day. It is reported consistently but does not appear in official documentation visible to a logged-out visitor, so treat it as reported rather than confirmed.
Signing up through our link: RunningHub's homepage states that an invite grants RH Coins to both the inviter and the new user, publicly quoted at 500 each. AIwithKay readers registering through the link on this page receive additional signup credits under our arrangement with RunningHub. We cannot independently verify the exact figure, because RunningHub publishes no referral terms to logged-out visitors, so check the balance credited to your account on signup.
What is NOT public: Subscription and membership prices. The plan page at runninghub.ai/vip-rights/2 opens a login wall, and both runninghub.ai/pricing and runninghub.ai/subscribe return 401. You cannot see what membership costs without an account.
The RH Coin to dollar conversion is not published anywhere. Without it, a rate of 0.2 RH Coins per second cannot be converted into real money, so we cannot show you the arithmetic. We are not going to guess at it.
Payments are non-refundable by default. The Service Agreement states that once the service is activated it is non-refundable unless severe defects render it unusable, that ending a plan early earns no refund for the remaining period, and that purchased coins expire after one year with no refund.
Pros
No install, no GPU: The single biggest barrier to ComfyUI is getting it running locally. RunningHub removes that entirely.
Genuinely current model catalogue: Recent models go live quickly, including Nano Banana 2.1, Seedance 2.5, Vidu Q4 Preview and MiniMax H3.
Video depth: Roughly 322 video models is far beyond what most cloud generation platforms offer in one place.
API pricing is published in plain dollars: The developer API lists per-second, per-call, per-image and per-million-token rates openly, with no account required.
Per-second billing: You pay for compute you actually use rather than an idle server, and the GPU tier rates are stated on app pages.
Real open-source presence: The HM-RunningHub GitHub account publishes maintained ComfyUI plugins with meaningful adoption, which is more transparency than most closed platforms offer.
A genuine free daily allowance: Reported consistently at 100 RH Coins credited on each daily sign-in, so you can keep using it at a small scale without ever paying.
Cons
You cannot see what a subscription costs before signing up: The plan page opens a login wall, and the pricing and subscribe URLs both return 401. That is a real problem for anyone trying to budget.
RH Coins have no published dollar value: Compute is billed at 0.2 to 0.4 RH Coins per second, but nothing public states what a coin costs, so platform compute is impossible to price in advance.
The official API Pricing page is empty: The link in RunningHub's own footer displays No Data under both of its sections.
Non-refundable by default: The Service Agreement states that once a service is activated it is non-refundable, and cancelling mid-term earns no refund for the unused remainder.
Purchased coins expire: Separately purchased RH Coins are valid for one year, after which the agreement states no refund is given.
ComfyUI is still ComfyUI: Hosting it in the cloud removes the install problem, not the learning curve. Node graphs remain intimidating if you have never used them.
The operator is disclosed only in the fine print: Nothing on the homepage says who runs RunningHub. The legal documents name Hong Kong Huize Technology Co., Limited, with the agreement governed by the laws of the People's Republic of China and disputes heard in Chaoyang District, Beijing.
The free allowance is not officially documented: The daily coin grant is reported consistently by third parties and in community posts on RunningHub's own site, but it does not appear in any official page we could reach without an account.
What Makes It Unique
Most AI platforms give you their own models behind their own interface. RunningHub instead hosts the open ComfyUI ecosystem in the cloud and bolts a 400-model aggregation API onto it, so one account covers both hand-built node workflows and one-call access to commercial video and image models. That combination of open workflow tooling and broad commercial model access in one place is the thing competitors generally do not offer.
Kay Score
6.5
/ 10
