OFICIAL Databricks Newsroom

Smart Routing in Unity AI Gateway: Match frontier quality with 30%+ lower cost per task

What happened
Based on Databricks Newsroom · Aug 13, 2026

Databricks introduces Smart Routing in Unity AI Gateway to automatically match coding tasks to the most cost-effective AI models based on complexity, reducing costs by over 30% while maintaining performance.

🚩 Report a correction ↗ Open source
Smart Routing in Unity AI Gateway: Match frontier quality with 30%+ lower cost per task
Databricks Newsroom — Databricks
Key points
·
In its prior post about benchmarking against the Databricks codebase, we found that models cluster into capability tiers and that much everyday work (e.g., flipping a flag, a single-file edit, a well-scoped bug fix) did not require the most expensive models.
·
So how do you reduce AI coding costs without sacrificing developer productivity?
·
One of the biggest opportunities is matching each task to the right model instead of defaulting every task to the most capable (and most expensive) option.
·
Just leveraging lower cost models can save you 50%+, but it’s incredibly daunting for users.
Key numbers
·
Early testing shows significant savings: internal workloads achieved 35% cost reductions, while public benchmarks demonstrated 56% savings compared to using a single high-end model.
·
By automating model selection, the feature helps organizations achieve frontier-level performance at a fraction of the cost, with savings exceeding 30% in tested scenarios.
·
Databricks introduces Smart Routing in Unity AI Gateway to automatically match coding tasks to the most cost-effective AI models based on complexity, reducing costs by over 30% while maintaining performance.

Databricks reports that coding tasks vary widely in complexity, with many routine edits not requiring the most advanced models. The company’s prior benchmarking showed that simpler tasks could be handled by less expensive models without sacrificing quality. To address this, Databricks developed Smart Routing, a feature in Unity AI Gateway that automatically selects the appropriate model for each task based on factors like complexity and cost. The system aims to reduce expenses while preserving developer productivity by avoiding the need for manual model selection or restrictive spending limits.

Smart Routing operates within tools like Claude Code and Codex, optimizing costs for existing workflows. It also integrates with Omnigent, a meta-harness that extends intelligent routing across both models and coding harnesses. Early testing shows significant savings: internal workloads achieved 35% cost reductions, while public benchmarks demonstrated 56% savings compared to using a single high-end model. The system uses a lightweight classifier to label tasks by complexity and language before routing them to the most suitable model, balancing performance and expense.

The feature is designed to handle real-world coding sessions, which often involve nuanced decisions across multiple sub-tasks. Smart Routing supports dynamic adjustments, such as delegating large codebase summarization to cheaper models while reserving advanced models for architectural design. This approach ensures that frontier models remain available for tasks that genuinely require them, optimizing both cost and productivity. Databricks emphasizes the importance of continuous feedback to refine routing decisions, logging session traces for evaluation while maintaining strict data governance standards.

Smart Routing is now available in Beta through Unity AI Gateway, offering teams a way to reduce AI coding costs without sacrificing developer choice or efficiency. By automating model selection, the feature helps organizations achieve frontier-level performance at a fraction of the cost, with savings exceeding 30% in tested scenarios. For teams seeking to balance performance and expense, Smart Routing provides an alternative to manual model selection or rigid spending controls.

Original source → Deals on Clipraptor.com →