Free GPU & Compute programs are offers that give you no-cost access to GPU or CPU resources for machine learning. You’ll usually get a hosted notebook (Jupyter-style), a specific GPU type when available (often T4-class or older GPUs), and usage limits such as daily hours, weekly quotas, or short session windows. Some programs are credits (a dollar amount applied to compute), while others are time-based. They’re best for prototyping, learning, small fine-tunes, evaluations, and running occasional inference without paying upfront.

This category includes programs from RunPod, Google (Colab and Kaggle), AWS (SageMaker Studio Lab), DigitalOcean/Paperspace, Modal, Lightning AI, Lepton AI, NVIDIA, and Lambda Labs. Each provider sets its own limits, GPU availability, and eligibility rules, so it’s worth comparing quotas and session constraints before you start.

Sometimes. Many “free tier” quotas are tied to your account and can’t be transferred, but some credit-based grants or promo balances may be transferable depending on provider rules. AI Credit Mart helps you find eligible listings and safely buy or sell unused AI credits when transfers are allowed.

Yes. Some offers are structured as bonuses rather than pure free tiers. For example, RunPod’s new-user bonus is awarded after you spend about $10 on the platform, and then you receive a random credit bonus (roughly $5 to $500) you can apply to services. Other options like AWS SageMaker Studio Lab are free to request and use, but may involve approval wait times and strict daily GPU limits.