What provider or pricing field should gpu.fund add first?

I am launching gpu.fund tonight and using the prelaunch window to sanity check the roadmap.

The simple version: it tracks live cloud GPU rental prices so people can compare H100, A100, RTX 4090, 5090, MI300X, B200, and similar options without opening a dozen pricing pages.

The messy part is deciding what to normalize next. Hourly GPU price is useful, but it is not the whole bill.

What would make this more useful for you?

- More providers

- Storage and bandwidth costs

- Interruptible or spot pricing

- Multi-GPU bundle pricing

- VRAM and region filters

- Availability history

- Inference vs fine-tuning recommendations

Current cache is 146 prices across 12 providers. I am especially interested in missing providers and hidden fees that make a cheap listing less cheap in practice.