Which tool do I need?

Last updated: September 2026

One rule each. Find your job below: every answer is free, account-less, and deletes your files automatically. If your text has the right content but tired wording, tool paraphrasing is the lighter fix: it keeps the facts, names and numbers and changes only the words and sentence shapes.

This is the plain answer to choose the right tool for the job and choose the right tool for the right job on this suite specifically: match your task to the row below rather than guessing between similar-sounding names. A few of these, PicReviver's restoration, BGNinja's cutout, genuinely run on a model rather than a fixed filter, so how to choose the right ai tool here comes down to whether the job needs that kind of pattern-matching or just a straightforward format change. What is the best way to choose the right ai tool from this list specifically: pick by what the file needs done to it, not by which tool sounds the most advanced.

The order you chain them in changes the result

Most jobs need two or three tools, and the sequence isn't arbitrary. A few orders are reliably better, for reasons that are worth knowing once:

• Restore before you cut out. A background remover decides where the subject ends by looking at edges. Sharpen and clean the photo first and those edges are clearer, so the cutout is more accurate. Do it the other way round and you're restoring an already-committed edge.
• Clean the audio before you transcribe it. Speech recognition loses accuracy on a noisy recording, and no amount of correcting the text afterwards recovers a word the model never heard. Noise removal first, transcription second.
• Separate stems before transcribing a song. Lyrics come out far better when you first extract vocals from song mixes, because the instruments aren't competing with the voice.
• Strip the metadata last. Any tool that saves a new file can write new metadata into it: a software tag at minimum. If privacy is the point, make the strip the final step, after every edit is done.
• Convert the format last, too. Each lossy save costs a little detail, so do your editing in the best format you have and convert once, at the end, to whatever the destination wants.
• Upscale at the end. Enlarging first means every later step works on four times the pixels for no benefit, and takes four times as long.

Three chains that come up constantly: old family photo → restore → remove date stamp from photo → strip metadata → print. Product photo for a listing → remove the background → convert to JPG at the size the site wants → strip metadata. Interview recording → remove noise → transcribe → edit the text.

Related: Discord Toolkit: Emoji Sizes, GIFs, Clip Cleanup, Does Resizing a Photo Remove Its GPS? Nine Operations Tested, I Tested the Metadata Remover on JPG, PNG and WebP.

Limits, at a glance

Every tool is free with no account. Results are deleted automatically within 45 minutes: on StemGrab, ScribeGrab and PassportNinja a pass stretches that to 24 hours, elsewhere 45 minutes is the rule on both tiers. What differs is how much you can put in at once: these are the free limits, and an optional pass (€5 for 30 days, €3 for 24 hours) raises most of them. Where a given tool needs a different number on its own page, that page is the accurate one: this list only aims for the quick answer:

• Metadata Remover, 60 MB per photo. JPG, PNG, WebP and HEIC; TIFF is deliberately refused rather than half-cleaned.
• PicReviver, 99 MB per photo, 10 photos per job. Input up to 2000 px on the longest side, output up to 8000 px.
• BGNinja, 99 MB and 30 megapixels per image, 3 images per batch.
• PassportNinja, 99 MB per photo, 6 photos per batch.
• ScanReviver, 30 pages per scan and 10 scans a day; the separate PDF tools do 25 pages and 250 MB per file.
• ConvertNinja, image conversion runs inside your own browser, so there is no server limit and nothing is uploaded. Video (1200 MB, 20 minutes), RAW photos (200 MB) and e-books (100 MB) do go through the server.
• StemGrab, 15 minutes of audio, up to 2 GB.
• ScribeGrab, 60 minutes of audio or video, up to 1.5 GB.
• CleanGrab, 45 minutes of audio or video, up to 1 GB.
• VideoBGNinja, 90 seconds of video, up to 2 GB.

What none of these tools do

Worth being straight about, so you don't spend twenty minutes looking for a feature that isn't here:

• No generating images from a prompt. Everything here works on a file you already have, with two exceptions: PicReviver's text to 3D model starts from a short description, and ScribeGrab reads typed text aloud, which is the mirror image of transcribing.
• No downloading from Spotify or any other platform: with one exception. ScribeGrab takes a YouTube link and transcribes it, because a subtitle track is often already there. Everywhere else: bring the file, the tools don't fetch it for you.
• No face swapping and no "make this person do something". Face restoration repairs the face that's in the photo, and stops there.
• No handwriting recognition. Text recognition handles printed and typed pages well and handwriting badly, which is a property of the technology rather than a setting.
• No video editor with a timeline. ConvertNinja does the single-step jobs, trim, crop, rotate, merge, change the speed, swap the audio, and CleanGrab cleans the sound inside a video while VideoBGNinja removes its background. What you won't find is a timeline, titles or transitions.
• No coding assistant. If you are looking for an agentic AI coding tool that writes and edits software for you, this suite does not have one: the text tools here rework prose you already have, not code.
• No file storage. Results are wiped within 45 minutes by design: 24 hours on StemGrab, ScribeGrab and PassportNinja if you hold a pass. Download what you need before you close the tab.
• No API platform. These are tools for a job in front of you, not a pipeline; the two exceptions are the free PDF API on ScanReviver and the background removal API on BGNinja. Batches are capped per tool: mostly 3 to 25 files at a time, though PicReviver goes to 1000 with a pass.
• Nothing that pretends a limit isn't there. Where a result has a ceiling: a very reverberant room, a watermark across the middle of a busy photo, a photo too small to enlarge usefully, the guide pages say so rather than promising a miracle.

Combining tools

The tools chain nicely: restore an old photo with PicReviver, cut it out with BGNinja, and strip the file's metadata here before posting. Or split a song with StemGrab and transcribe the isolated vocal with ScribeGrab. Every step is free, so experiment away. Whichever combination you land on, use the right tools for each step, restore before you cut out, not after, rather than one tool trying to do all of it. Two of the text tools rewrite wording and they split the job: the Paraphrasing Tool keeps your tone and changes the words, while the AI humanizer tool also changes the style, so pick by whether the tone should stay.

Why the batch caps are the shape they are

The numbers above aren't round marketing figures; each one traces back to a queue on our own GPUs. Restoration and background removal are GPU-bound, so a batch cap on PicReviver or BGNinja is really a promise about how long the queue behind you stays short: raise it and everyone's wait grows, including yours on the next job. StemGrab and ScribeGrab are bound by a different resource: the length of the recording, because separating stems or transcribing scales with duration rather than file size, which is why their limits are stated in minutes and not megabytes. ConvertNinja is the odd one out on purpose: image conversion runs inside your own browser tab, so there's no server limit to set, and the cap you'll actually hit is whatever your phone or laptop can hold in memory. When a limit changes, it's because the hardware changed or the queue got measurably worse, not because free was made to feel smaller. If a job of yours keeps landing right at a limit, that's usually a sign the paid pass fits that one tool, not a sign the free tier is deliberately stingy.

Browse by type instead: the photo tools, video tools, audio tools, transcription and subtitle tools, PDF tools, file converters and text tools each have a hub page that lists every tool of that kind.