PINVOU / Download & Connect
From client installation
to a working model connection
Install PINVOU on Windows, macOS or MegaCube, then connect to a deployed model service to start working with your files and tasks.
Install PINVOU first,
then download models within PINVOU
The PINVOU client includes Model Store, conversations, tools and workflows for creating files. Model downloads are now managed within PINVOU; a separate local model management installer is no longer needed.
- 01
Check MegaCube
Make sure the device is powered on and storage and networking work. Confirm how many devices your target model needs and any other deployment requirements.
View validation records → - 02
Choose a PINVOU Version
Choose the Windows, macOS or MegaCube/Linux client for the computer running PINVOU.
Choose your client → - 03
Download models in PINVOU
Launch PINVOU, open Model Store, download and start a model, then begin a conversation.
View Model Store steps →
Choose Your PINVOU Client
Download directly from pinvou.com without visiting another website. These are the original H3C 0.10.4 installers, mirrored on October 15, 2026. Check your operating system and chip architecture before downloading.
x64 · EXE installer
0.10.4 · 153.5 MiB
Universal · DMG installer
0.10.4 · 195.9 MiB
ARM64 · DEB installer
0.10.4 · 130.4 MiB
Read and follow the H3C software license before downloading or using the software. The source pages state that these packages are for use outside the United States. Download SHA-256 checksums ↓
Package sources and release information
The official files are mirrored without repackaging. SHA-256 values were computed when mirrored to verify download integrity. Official release information: Windows · macOS · MegaCube / Linux. Release channels may differ; GitHub Releases may include preview builds separate from these official installers.
Check These Four Items Before Connecting
Once the device, model, network and test task are ready, verify your first connection.
MegaCube is powered on and its storage, memory and operating system are working properly.
The target model is deployed and the inference service is running.
The computer running PINVOU can reach the service address and port.
Send a short question and confirm that the selected model responds.