TikTok Caption Generator (FREE TOOL)

A TikTok caption is the post text: the words, hashtags and mentions under the video, next to the account name. Subtitles, which TikTok also calls captions, are the words on top of the video that follow what is said.

This tool writes the first kind. It reads what the video shows and what you sell, and writes the text a viewer reads before they decide to watch, comment or tap the button. For on-screen subtitles you need a transcription tool that listens to the audio, which is a different job.

The two meet in one place: TikTok's safe zone. The caption covers the bottom of the video, so keep text in the picture out of that area. The TikTok safe zone checker shows what the caption and the buttons cover.

TikTok counts hashtags inside the caption: the Content Posting API says hashtags and mentions in the caption are matched, and the 2,200 character limit covers the whole text. TikTok does not publish a best number of hashtags.

This generator writes 3 to 5 hashtags for a post, 1 or 2 for a Spark Ad and none for an in-feed ad. Each one names the product, the brand or the niche, so it describes this video. It never writes #fyp, #foryou or #viral: a tag that fits every video says nothing about yours.

Want only hashtags? Run the tool, then copy the last line of the caption you like. For a post, that line is the hashtags alone.

Last checked: 21 October 2026. The post limit comes from TikTok's Content Posting API reference; the 4 lines, the 100 character ad text and the 2 character emoji from the TopView specs; the no hashtag rule for in-feed ads from the in-feed ad specs, each read at the source.

• The count was worked by hand on seven cases, including an emoji, an accented letter and full-width characters in an ad, and the tool and the page agree with every one.
• Four fit cases were worked by hand: a post, a Spark Ad that keeps its hashtag, a Spark Ad that drops it to stay under 100, and an in-feed ad that loses every # and @.
• The code that assembles each caption strips em and en dashes, links and generic tags, and sends a caption back once when its hook announces the video or it breaks a limit.
• Typical, boundary and hostile inputs ran through the real page with the model reply stubbed: an empty form, a one line video, a non-English video and an instruction typed into the video box.
• The eight frames above are Selzee gallery ads with their own lines as captions, laid out by the same placement rules the tool uses.