Token Budget

The planned or enforced limit on how many tokens an agent is allowed to use across a task — balancing capability (more tokens = more thinking) against cost and latency.

📌 Example: Setting a max of 10 LLM calls per task, each with a 4k token output limit, keeps costs predictable for a high-volume agent.

🛠 In practice: Token budgets force thoughtful architecture. Agents that can run indefinitely are expensive and often get stuck. Budgets create pressure to design efficient prompts and early stopping conditions.

• constrains Agentic LoopA token budget sets an upper bound on how many loop iterations or LLM calls the agent can make before it must stop.