Fig selecting the right model for a first-call deck request

Auto routing

Fig saves 81% in token costs.

Fig auto routing selects the right model for each request, with controls to optimize for cost efficiency, balanced performance, or frontier capability.

Get a demo

70%

Fewer tokens per task compared with Claude Cowork

81%

Lower token costs compared with Claude Cowork

78%

Preferred responses compared with Claude Cowork

Evaluations

Routing informed by enterprise tasks.

Fig evaluates models on the work enterprises actually do, from writing and research to analysis and creation, then routes by reasoning, quality, modality, complexity, cost, and task type.

Fig routing different enterprise tasks to models based on task requirements

Optimize routing

Routing that keeps enterprises in control.

Choose cost-efficient for everyday work, balanced for the ideal price-performance tradeoff, or frontier for the hardest tasks. Set the right mode by department while Fig selects the model request by request.

Cost-efficient, balanced, and frontier Fig routing modes assigned by department

AI usage controls

Routing that is visible.

See what is driving usage across people, departments, products, and models. Track actual and projected spend, then adjust routing controls before budgets become limits.

Explore usage controls →

Actual and projected AI usage cost trends across Fig Assistant, Agents, and MCP

Route every request with purpose.