1
Open your workspace
Sign in at platform.r3al.ai, then choose New project from Projects. You need an admin or developer role to create a project.
2
Answer the setup questions
Choose what the model should do and how demanding the work is. Enter how many people will use it at the same time, your preferred answer speed, and the amount of text in each request.Finally, choose whether you have your own infrastructure or want R3AL to host it. If you have GPUs, select their type and quantity. Use simultaneous AI users, not your total registered users, when estimating demand.
3
Review your recommended setup
Review the suggested model, precision, GPU configuration, and estimated performance. Compare any alternatives and use Back or your previous answers to adjust the requirements.These figures are sizing estimates to validate against your workload. Selecting a recommendation does not reserve GPUs or bring a model online.
4
Create your project
Give the project a name. Under Add details, describe your use case and what a good answer looks like. Select We already have a fine-tuned model only if you have one to bring. Choose Create project.
5
Finish your project brief
In Overview, follow Continue setup if shown. Review the task, data source, processing locations, expected outputs, and budget, then save the project plan.For a new model, use Data to prepare and review examples. If you are bringing a trained model, add its details under Models with your technical collaborator.
6
Arrange model setup
Discuss your pilot with R3AL and share the project name or link. Agree on the model, data requirements, hardware, and serving connection for your project. See Set up your model.