Open Model Archive compares AI models by what they build
Archive is a community benchmark that compares AI models through real outputs, not test scores. Different models receive the same task, then their unchanged results are placed side by side. The first examples include simple games and generated by different models.
Anyone can take part by copying the project , running the task with a chosen model, and sharing the generated result. The goal is to show what each model can actually create before a person fixes or improves it.
Key points
- Archive gives different AI models the same task.
- Models are compared by their actual generated outputs.
- Early examples include AI-made games and .
- The comparison uses raw output without human changes.
- Anyone can copy the , run a task, and contribute a result.