A common reliability pattern for long-horizon agents is to run an expensive critic/verifier model after every step - which roughly doubles token spend even though most steps are unambiguous and gain nothing from a second opinion. Instead, read two cheap uncertainty signals from the actor's own next-action distribution - global entropy and the top-2 logit margin - to approximate the Value of Information of a critique, and invoke the critic only on the genuinely ambiguous steps. Let the actor internalize the critic's guidance over the run so the calls taper off as it improves.
Call the Critic Only When the Agent Is Unsure: Gate Verification on Action Uncertainty
🔒 Pro tip · Advanced
Unlock this tip — and 140 more
This is one of 141 advanced, fact-checked tactics reserved for Pro. Get the full 163-tip library, a searchable archive, and a new tip every morning. Free for 7 days, then $9/mo.
Prefer to browse? The 22 Beginner tips are free forever.
More in Model Selection
🎚️Model Selection
60-80% on routed traffic
Stop Paying Frontier Prices for Boilerplate Work
Most of your token spend is on tasks a small model handles perfectly. Match the model to the job instead of defaulting to your most expensive option for everything.
🎚️Model Selection
Often cheaper than escalating a tier on reasoning-limited hard tasks (qualitative — depends on the price gap and how many extra reasoning tokens the cheaper model spends)
Buy More Reasoning on the Cheap Model Before You Upgrade the Tier
When a cheap model stumbles on a hard task, the reflex is to jump to the frontier tier. Often the cheaper move is to keep the small model and turn its reasoning effort up — its per-token rate is so low it can brute-reason through the problem and still cost far less.
🎚️Model Selection
40-70% when most queries are easy
Cascade: Try the Cheap Model First, Escalate Only When It Fails
Send every request to a small model first, programmatically check the answer, and only escalate to a frontier model when the cheap one falls short.