How to summarize a YouTube video with ChatGPT

Key Takeaways

• ChatGPT cannot watch a YouTube video. If you paste a video link, it typically summarizes the title and description — not the actual content.
• The reliable method is a transcript: get the video transcript, paste it into ChatGPT with a summary prompt, and it works well.
• Very long videos can exceed what a single chat handles cleanly — split the transcript or ask for a section-by-section summary.
• Total effort is 3–5 minutes per video across two tabs; fine occasionally, tedious as a routine.
• If you summarize YouTube videos regularly, a one-click tool like Zloth does transcript + summary + saving directly on the thumbnail.

Can ChatGPT summarize YouTube videos?

Not directly — and this is the part most guides skip.

ChatGPT cannot watch or listen to a video. If you paste a YouTube link into the chat, it reads the video's page at best: title, description, maybe some comments. The "summary" you get back is often a plausible-sounding guess.

What ChatGPT does brilliantly is summarize text. So the working method is simple: give it the video's transcript.

The 4-step workflow

1. Get the video's transcript. Paste the video URL into our free YouTube transcript generator and copy the result. (Alternative: YouTube's own transcript panel — see how to get a transcript of a YouTube video.)
2. Open ChatGPT and start a new chat.
3. Paste this prompt, then the transcript:
   Summarize this YouTube video transcript. Give me: 1) a 3-sentence overview, 2) the key points as bullets with timestamps, 3) any actionable advice. Transcript: [paste here]
4. Refine with follow-ups. "Expand on the point at 12:40", "What did they say about pricing?", "Turn this into 5 tweet-length takeaways."

Expect 3–5 minutes per video across two tabs. That is genuinely fine for the occasional video — and genuinely tedious if you do it every day.

Better prompts for specific outcomes

The generic prompt above works, but ChatGPT gets noticeably better when you tell it what the summary is for. Swap in one of these depending on the job:

One habit worth stealing: keep the timestamps in the transcript when you paste it. Every prompt above gets more useful when the answer can point you to the exact minute to verify or rewatch.

Long videos: the split-and-merge workflow

A talking-head video produces roughly 7,000–9,000 words of transcript per hour. One hour usually fits in a single ChatGPT message; a 2–3 hour podcast or lecture often doesn't — the model starts ignoring the middle or the reply gets vague. The fix is mechanical:

1. Split the transcript at natural breaks — chapter marks or round timestamps. Two to four parts is almost always enough; keep each part under ~8,000 words.
2. Summarize each part in the same chat: "This is part 1 of 3 of one video's transcript. Summarize it with timestamps; I'll send the next part after."
3. Merge at the end: "Now combine the three part summaries into one summary of the whole video: overview, key points with timestamps, actionable advice."

Budget 10–15 minutes for a long video. This is the point where the manual workflow stops being "free" in any meaningful sense — fine for a one-off deep dive, painful as a routine.

Troubleshooting: when the summary comes out wrong

• The summary sounds plausible but wrong: you probably pasted a link instead of a transcript. Test it — ask about a specific detail from the middle of the video. If ChatGPT can't answer, it never saw the content. Paste the transcript text itself.
• Names and technical terms are garbled: that's the auto-generated captions, not ChatGPT. Add a line to your prompt: "The transcript is auto-generated; fix obvious speech-to-text errors in names and technical terms from context."
• The reply ignores the end of the video: the transcript was too long for one message. Use the split-and-merge workflow above.
• The video is in another language: paste the transcript as-is and ask for the summary in English (or any language) — translation plus summarization in one step is something ChatGPT handles well.
• No transcript available at all: the video has no caption track (common for music, brand-new uploads, and some languages). No tool can conjure one — see what to do when a video has no captions.

The limits to know about

• Long videos: a 1–2 hour video produces a transcript that can overwhelm a single chat. Split it into parts, summarize each, then merge the summaries.
• No captions, no luck: if a video has no caption track at all, there is no transcript to paste. That is a YouTube limitation every tool shares.
• Auto-captions are imperfect: auto-generated transcripts mangle names and jargon; the summary inherits those errors.
• Nothing is saved: the summary lives in your chat history. Finding "that video summary from three weeks ago" means scrolling.

Honest verdict: ChatGPT is a perfectly good free option for occasional summaries. Extensions like Glasp's "YouTube Summary with ChatGPT" semi-automate the copy-paste. The step beyond that is skipping the workflow entirely.