Would You Use a Real-Time Auction Marketplace for AI Inference?

We're building infer.bid, a real-time matchmaking platform that connects suppliers and consumers of AI inference compute. Our primary focus is on open-source models initially, aiming to tackle high inference costs and GPU scarcity by introducing dynamic pricing through a real-time bidding system.

We’d love your input:

If you use services like OpenRouter, Replicate, or Hugging Face, what are your biggest pain points? Is cost your primary issue, or are you looking for more advanced features such as intelligent model routing, caching, or prompt optimization to save costs?

Would a platform that lets you dynamically bid for inference compute be compelling for your use cases? How important is pricing transparency and the ability to choose your hardware provider directly?

We're trying to better understand different user scenarios and prioritize features that directly address the community's needs. Any insights or feedback would be greatly appreciated!

Thank you!

We've got xena's Skypilot already, how would you differentiate your offering? https://xeiaso.net/talks/2025/ai-chatbot-friends/