Inference Cost Tracking And Budget Enforcement With Per User Quotas

via “budget and cost management with token tracking and rate limiting”

Open-source infrastructure for Computer-Use Agents. Sandboxes, SDKs, and benchmarks to train and evaluate AI agents that can control full desktops (macOS, Linux, Windows).

Unique: Implements a budget management system that tracks token consumption and costs across heterogeneous VLM providers with provider-specific pricing models, supporting per-agent/per-task/global budget constraints with automatic throttling or termination. Integrates with provider APIs for real-time cost tracking.

vs others: More comprehensive than simple token counting because it tracks actual costs across providers with different pricing models; automatic throttling prevents budget overruns vs. requiring manual monitoring.