Baidu PaddlePaddle AI Studio Daily Free GPU Compute (8 points / day, V100)

Running a project each day earns 8 compute-card points; 1 point buys 2 hours of V100 16GB compute or 1 hour of V100 32GB compute, i.e. roughly 16 hours of V100 16GB per day; a beginner task adds 100 more points, and each referral boost adds 10 points

Official help doc "Compute cards" verbatim: GPU compute on AI Studio comes mainly in three ways: automatically earned by running projects, earned by sharing and inviting new users, and issued manually by operations staff. The compute card unit is the point; each point buys 2 hours of high-end GPU V100 16GB compute or 1 hour of high-end GPU V100 32GB compute. Running a project each day earns 8 points, which can be used for 16 hours of V100 16GB compute; a certain amount of compute-card time is granted each day when you click "Start environment"; every new user you pull in to boost earns you 10 points (the boosted user must have registered within the last 48 hours); once a boosted new user completes the beginner tasks, you earn another 100 points. Billing rules: billed per minute, anything under 1 minute is not billed, and charges are totaled every 6 minutes; details can be viewed on the "Compute card details" page, and issuance records under "Compute card management" in your personal center.

⚠️The official doc carries no update date, and the GPU types (V100 16GB / 32GB) are as stated on the official page; the platform has also launched A100 and domestic DCU resources in recent years, so the options you can actually pick are shown on the environment launch page. How long the daily free compute stays valid is not stated clearly on the official page; go by what "Compute card management" in your personal center shows. This is a Notebook-style cloud GPU, mainly for learning, training small models, and getting projects running; it is not a large-model inference API, and it is a separate benefit from the baidu-aistudio-llm-api entry on this site (the Token quota).