Hey folks! 👋 I keep seeing questions about Hailuo's pricing model pop up, especially around "credits." It can be a bit confusing when you're used to thinking in pure API calls from other services.
So, think of it like this: an **API call** is the action ("send this prompt"), while a **credit** is the currency needed to *pay* for that action. The key is that not all API calls cost the same number of credits.
Hailuo uses credits to normalize cost across different models and operations. For example:
* A simple completion with a smaller model might cost 1 credit per 1K tokens.
* The same call to a much larger, more powerful model might cost 8 credits per 1K tokens.
* An API call for image generation likely uses a totally different credit formula.
Here's a simplified analogy from our world: one Datadog APM span is an "API call," but the "credit" cost depends on if it's a nanosecond trace from a test or a 10-minute, high-cardinality trace from production.
They bundle credits in packs (like 1000 credits for $X). This lets them offer a single pricing sheet even as they add new models with different underlying costs. Always check their latest docs for the credit schedule for the specific model you're using!
Dashboards or it didn't happen.
That analogy to Datadog's tracing actually helps a lot, thanks. It makes the abstraction clearer.
If the credit cost varies per model, does that mean Hailuo's pricing page only lists the credit pack prices, and you have to dig into a separate model documentation page to find the actual cost per operation? That seems like an extra step for budgeting.