FAQ
Sumari is free. No paid features, no subscriptions. It uses a BYOK (Bring Your Own Key) model: you provide your own AI service API key, Sumari calls your account directly. No middleman, no markup. You only pay your AI provider.
YouTube, Bilibili, X/Twitter, Vimeo, Instagram, and other yt-dlp-compatible sites. You can also import the current video from your browser with one click, or drag local audio and video files in from Finder.
Yes. Sumari tries platform subtitles first. If none are available, it falls back to speech recognition: cloud transcription, local Whisper.cpp, or Apple Speech (macOS 26+).
Yes. Export to Markdown if you want to edit it, or PDF if you want to keep the layout and send it as-is.
For summaries: Apple Intelligence (macOS 26+) and any OpenAI-compatible API, including OpenAI, DeepSeek, Groq, Ollama (local), and others. For transcription: platform subtitles, cloud Whisper API, local Whisper.cpp, and Apple Speech. You can choose different providers for summaries and transcription.
It works, but the built-in Foundation Model has a 4,096-token context window (input + output combined). That caps out at very short videos. Longer content hits the limit and fails. It also requires macOS 26+ and may not be available in all regions or on all hardware. For normal-length videos, use OpenAI, DeepSeek, or another cloud provider.
Yes. Paste a YouTube or Bilibili playlist link, and Sumari expands it into individual videos and processes each one.
Sumari bundles yt-dlp, and yt-dlp itself supports 1,700+ sites. If a site is not on our tested list, paste the link and choose "Force Analyze." Sumari will let yt-dlp try. Most mainstream platforms work, but we do not guarantee every site.
macOS 14 (Sonoma) or later. Apple Silicon and Intel Macs both work (Intel needs a 2018 or newer model). Apple Speech and Apple Intelligence require macOS 26 (Tahoe). Apple Intelligence availability also depends on region and hardware.