01
¥8,000 · OpenBuild the first public inference benchmark for domestic AI PCs
Complete reproducible tests across three devices, five models and two engines.
Turn real needs into verifiable open tasks
Every bounty states deliverables, acceptance criteria, open license, reviewers, deadline and reward.
02
¥5,000 · 2 collaboratorsA reference mobile voice assistant
An end-to-end demo of offline ASR, a lightweight language model and TTS.
03
¥3,000 · OpenLicense data cleanup for Local AI open-source projects
Verify code and model licenses for 100 projects and flag commercial-use boundaries.
04
Joint bountyCall for real enterprise scenarios and bounty topics
The community helps break needs into open, reviewable, verifiable tasks.