A Powder.gg alternative
for a tool that stopped.

Seven years, and then a letter.

Visit powder.gg today and you do not get the product site. The domain redirects to a document addressed to the Powder community, signed by the team. In it they explain that they stretched their finances as far as they could and, in their words, "we can no longer sustain the app". From that day the app would no longer be updated or maintained. Active subscriptions were cancelled so they would not renew. For yearly plans they said they would do their best to refund the unused portion where possible, and apologised in advance for the cases where it would not work out.

We are not going to enjoy this. A team spent seven years building something people liked, ran out of runway, and shut it off in about the most decent way available: warn everyone, stop the billing, refund what you can, say thank you. Plenty of companies do considerably worse. The reason this page exists at all is that people lost a working routine overnight and are now searching for where to go, and the useful thing we can do is be accurate about it.

More than a clipper, which was part of the point.

Powder was a Windows desktop app, not a website. It shipped its own low-resource screen recorder, found highlights in your recordings and streams automatically, cut them to vertical templates, generated subtitles, and assembled montages. It supported a broad list of games and had a genuinely active community around it. If you are reading this because your workflow just disappeared, that is the shape of what needs replacing, and it is a wider shape than most of the tools now bidding for your attention.

One detail matters more than the rest, and it is the reason this page is not the easy sales pitch it looks like. Powder ran its analysis on your own machine. Their own site said the software was designed to run entirely locally on your Windows PC, precisely so they would not carry cloud compute bills for every hour of gameplay a user threw at it. Powder Pro went further in that direction, optimised for AMD hardware and its neural processing units for real-time work while you played.

So Powder was not a cloud service that got switched off. It was a local app, made by a company that ran out of money.

Local did not save Powder. Here is what it does do.

The convenient argument here writes itself: cloud tools can be taken away, so buy a local one. Powder does not support that argument, and pretending otherwise would be the kind of thing you should distrust a vendor for. Powder was local and it still ended. Anyone selling you "local" as insurance against a company failing is selling you something that does not exist.

What is true is narrower, and worth getting right, because it is the part that decides how badly a shutdown actually hits you.

• A locally installed app cannot be switched off from the outside. Nobody can revoke a copy that is already running on your machine, and there is no server that has to answer for it to open. That is the actual argument, and it is the whole argument.
• What stops is the future, not the present. When a company goes, you lose updates, fixes for the next Windows or game patch, support, and anything that reaches a server. You do not lose the ability to open what you installed and process a file that is already on your disk.
• Recurring billing is the thing that ends abruptly. A subscription is a relationship someone else can close. A one-time purchase has no renewal date to be cancelled on your behalf, which is why the Powder letter had to talk about cancellations and partial refunds at all.
• Your footage was never the hostage. This is the underrated one. Because Powder kept the analysis on your PC, nobody's shutdown took your recordings with it. Users of hosted tools whose libraries live on a vendor's servers have a harder morning when that vendor stops.

And the same test, pointed at us. CutBench is not exempt from any of this. It is a company too, and companies stop. Two parts of CutBench are not local: the license is activated through Steam, and publishing to TikTok or YouTube goes through those platforms' APIs, which change whether we are here or not. Those are the parts that would age if development stopped. The processing, which is the part you are actually buying, does not call anything: transcription, detection, ranking, trimming, captioning and rendering all run on your GPU, on files on your disk. That is what an installed copy would keep doing. We would rather set that expectation now than have you find the edges of it later.

Powder as it was, and what CutBench does.

Powder details are taken from its own site as archived by the Internet Archive on 15 March 2026, and from the team's farewell letter, both checked on 14 August 2026. We are deliberately not quoting Powder's prices: the pricing page went with the site, and a number we cannot re-check is not a number we will print. Blank cells mean the feature was not offered, not that it was offered badly. "Not verified" means exactly that, and we left it rather than guess.

Read the hardware line before anything else.

• On AMD, Intel, a Mac or a laptop? Go hosted. Browser-based clippers run the work on their servers, so the machine you own stops mattering. You will be paying monthly and uploading your footage, which are real costs, but they are costs you can actually pay. We compare three of them honestly: StreamLadder, Eklipse and FragCut.
• Want the recording part back? No clipper we know of replaces Powder's built-in recorder one-for-one. Most people end up pairing a separate capture tool with a clipper. CutBench takes files, so whatever you record with is fine.
• On NVIDIA, and it was the local part you liked? That is the case CutBench is built for. Same principle Powder ran on, keep the footage and the compute on your own machine, with a one-time price instead of a subscription. Start with the free demo, on your own recordings, before paying anything.
• Whatever you pick, keep your files. The clips Powder made are ordinary video files on your disk and they are yours. Nothing about the shutdown reaches them. If any tool you move to keeps your library on its servers instead, that is worth noticing now rather than the next time this happens to somebody.

One session in, a ranked list out.

You point CutBench at a full recording and walk away. It transcribes the speech, detects audio events like gunfire and explosions, tracks scene changes, and runs a trained classifier over the HUD that recognises the wounded screen, the death screen, the inventory, the map and farming. Those signals are scored together, each candidate is trimmed to its payoff, cropped to 9:16, captioned and labelled with a hook, and you come back to a ranked list rather than a timeline.

Worth stating plainly, because the category is loose about this: there is no text recognition anywhere in the pipeline. Nothing reads a kill feed, a scoreboard or any text on screen. The ranking is tuned on Rust and ARC Raiders, with a generic profile everywhere else, and there is a fuller account on the Rust and ARC Raiders pages.

• Windows 10 or 11, 8 GB of RAM and 16 GB recommended, 30 GB of free disk, and an NVIDIA GPU with 6 GB of VRAM and CUDA 11.8 or newer, so an RTX 2060, RTX 3050 8GB or RTX 4050 and up.
• Roughly 2 to 5 hours to process a 3-hour session on an RTX 3060 class card. It is a queue you start and leave, not a tab you sit in front of.
• Nothing uploads. The recording stays on the drive it was made on, party voice and all, and there is no account to create before you can process a session.
• $34.99 once on Steam, with a free demo that runs the identical pipeline and marks the exports with a watermark.
• Offline after the first run. The models ship inside the build, so it works with no connection from the start.

Powder-refugee FAQ.