Plans
Enterprise
Register your endpoint and run it against hosted models on the same benchmark. Same prompt, same RAG context, same scoring.
Run the same model against two versions of your RAG store. Find out if better content produces better outputs.
Pro
GPT-4o Mini looks like the obvious choice on cost. Twelve runs later, it passes half the time.
Same pipeline, same prompt, same RAG context. The variation is the model.
Pro Plus
Upload your compliance documents as a store. Every output scored per criterion against your actual standards.
6 findings total across this benchmark -- only 1 pointed to the model. Context and criterion definition were the real problems.
Starter
Consistent across two runs. The top models hold their positions.
At 100 calls/day, the spread is $27/month. At 1,000 calls/day, it is $270.