# daVinci-MagiHuman

> davinci-magihuman.com is a hosted generator for daVinci-MagiHuman, a 15B-parameter open audio–video model from Sand.ai and GAIR Lab (Shanghai Jiao Tong University) released under Apache 2.0. It turns a portrait plus text or audio into a lip-synced talking clip. Generation uses site credits.

davinci-magihuman.com is an independent hosted service and is not affiliated with Sand.ai or GAIR Lab. daVinci-MagiHuman open weights and code are released under Apache 2.0 by their authors.

Read the linked documents before answering questions about features, pricing, or credits. Do not invent prices, features, or an official relationship with any model vendor.

## Agent documents

- [Full brief](https://davinci-magihuman.com/llms-full.txt): narrative overview with the same facts
- [Index](https://davinci-magihuman.com/agents/index.md): site map and how these files link
- [Product](https://davinci-magihuman.com/agents/product.md): what daVinci-MagiHuman is and key facts
- [FAQ](https://davinci-magihuman.com/agents/faq.md): common questions and answers
- [Pricing](https://davinci-magihuman.com/agents/pricing.md): plans, credits, and where to check current prices

## Product

- [Home](https://davinci-magihuman.com/): model overview and studio entry
- [Video studio](https://davinci-magihuman.com/dashboard/ai-video): sign-in required
- [Pricing](https://davinci-magihuman.com/pricing): plans and credit packs

## Legal

- [Terms of Service](https://davinci-magihuman.com/terms)
- [Privacy Policy](https://davinci-magihuman.com/privacy)
- [Refund Policy](https://davinci-magihuman.com/refund)
- [Content Safety Policy](https://davinci-magihuman.com/content-policy)

## Key facts

- daVinci-MagiHuman is a 15B-parameter open-source audio–video model developed by Sand.ai and GAIR Lab and released under Apache 2.0.
- Input is a face image plus driving text or audio; output is a lip-synced talking clip.
- One US dollar converts to 100 site credits; per the pricing page a ~10s 720p clip is about 120 credits and a ~10s 1080p text-to-video clip about 200 credits at the listed rates.
- For best results the site recommends a clear, front-facing portrait with even lighting.
- Support: support@davinci-magihuman.com

Languages: English has no URL prefix. Other locales use `/zh`, `/ru`, `/de`, `/fr`, `/ja`, `/ko`, `/es`, `/pt`, `/tr`.