BAGEN: Are LLM Agents Budget-Aware?
About
While agents are increasingly spending more resources, today agent cost is mostly measured only after execution. A Budget-Aware Agent (BAGEN) should treat budget as an active control signal, rather than a passive cost metric. We first systematically define budget estimation as internal budgets (from agent computation) and external budgets (from agent actions). We then formalize budget-awareness as progressive interval estimation: at each step of a plan, an agent should predict an upper and lower bound on remaining budget, and alert when completion is unlikely. Scoring with a rollout-replay protocol, we find consistent failure patterns on four environments and five frontier agents: (1) strong agents do not necessarily have strong budget-awareness, with correlation r=0.35. (2) frontier models are consistently over-optimistic, continue spending on tasks that are unlikely to succeed, instead of alerting the user early. (3) budget-aware signal is actionable and trainable. Early stop saves 28-64% tokens on failed trajectories, and SFT+RL strengthens early stop and alert behavior. (4) precise interval calibration remains challenging, with interval coverage capping at 47% after SFT+RL. Project page: https://ragen-ai.github.io/bagen/
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Early-stopping budget estimation | Search-R1, Sokoban, SWE-bench, and Warehouse Aggregate (test) | -- | 5 | |
| Feasibility Prediction | SWE-Bench | -- | 5 | |
| Feasibility Prediction | Search-R1 | -- | 5 | |
| Feasibility Prediction | Sokoban | -- | 5 | |
| Feasibility Prediction | Warehouse | -- | 5 | |
| Interval Quality | SWE-Bench | -- | 5 | |
| Interval Quality | Search-R1 | -- | 5 | |
| Interval Quality | Sokoban | -- | 5 | |
| Interval Quality | Warehouse | -- | 5 | |
| Task Performance | SWE-Bench | -- | 5 |