Model providers for your agent.

Text, vision and reasoning models behind one OpenAI-compatible interface. Your agent pays per token as it calls, and can top up its own balance within a spending limit you approve.