Video-analysis tool v0.5.1 adds direct AI analysis with Gemini, Kimi, OpenAI and Claude
sujingshen · x · 2026-08-04
A video-analysis tool has shipped v0.5.1 with a new AI direct analysis feature.
Users can enter their own API key once and then run full-film analysis directly, without the old ZIP upload / JSON copy / paste workflow. It supports preset providers including Gemini, Kimi, OpenAI, and Claude, and also allows custom OpenAI-compatible endpoints.
A few implementation details stand out:
- models that can see images will automatically receive frame collages
- analysis quality is said to match the older, uncompressed workflow
- the API key stays only in the local browser
- DeepSeek is not preconfigured because text-only models cannot inspect visuals well; users can still enable it manually with collage disabled
- the online trial does not support this feature; it is only available in the local full version
More from Multimodal
- Finger Frame Live AI turns a hand frame into real-time stylized video with Decart — venturetwins · 2026-08-04
- Decart-powered app turns your face into different styles in real time — venturetwins · 2026-08-04
- VAD isolates visually supported corrections for multimodal distillation and beats direct teacher supervision on six benchmarks — Kangning Zhang · 2026-08-04
- MiniMax H3 generates a rock-eating Dwayne Johnson swarm in 20 minutes — Furginator · 2026-08-04
- Codex uses Blender and Unity to build an otter-mech game with full assets — emollick · 2026-08-04
- Minimax H3 turns a Will Smith spaghetti rap prompt into gibberish — blastbottles · 2026-08-04