Advisor

This pattern fits long-horizon agentic workloads (coding agents, computer use, multistep research pipelines) where most turns are mechanical but having an excellent plan is crucial.

Agents that use a single model face a tradeoff. A fast, low-cost model keeps routine steps quick and less resource intensive, but may be less well suited for thorough analysis and decisions on complex tasks.

The agent runs on a fast executor model, and when it reaches a decision it cannot resolve confidently, it calls the Model Consult tool to get guidance from a stronger advisor model. The agent then continues the task itself, using its own tools.

That is what an Advisor is for: guidance from the stronger model, decision left with the Agent.

Eve Legal, quoted by Anthropic

On structured document extraction tasks, the advisor tool enables Haiku 4.5 to dynamically scale intelligence by consulting Opus 4.6 as complexity demands, matching frontier-model quality at 5× lower cost.

Genspark, quoted by Anthropic

We saw clear improvements in agent turns, tool calls, and overall score — better than a planning tool we built ourselves.

Claude Code

The advisor tool lets Claude consult a second, typically stronger model at key moments during a task, such as before committing to an approach, when stuck on a recurring error, or before declaring a task complete.

It adds less value on short tasks where there is little to plan, or on work where every turn needs the strongest model.

An Agent where every step is the hard step should just run on the stronger model, and one that calls the Advisor every turn has bought the expensive model with extra latency attached.