Cursor introduces Router for intelligent model routing in Auto mode
TL;DR
Cursor Router drives Auto mode by selecting optimal models based on task complexity. It offers Cost, Balance, and Intelligence optimization modes with admin controls.
What changed
Cursor Router now powers Auto mode by automatically picking models based on task complexity. The update adds Cost, Balance, and Intelligence optimization modes along with full admin controls. Vibe Builders, Basic Users, and Developers will see these routing decisions appear during normal sessions.
Why it matters
Basic Users gain easier access to suitable models for routine work without extra steps. Developers working on varied code tasks see the Balance mode handle typical project needs in ways that differ from fixed setups in tools like GitHub Copilot. Vibe Builders can rely on the Intelligence mode for more demanding prompts while keeping costs in view.
What to watch for
Test outputs against manual model picks inside Cursor to verify the router choices. Developers should run identical tasks in Cost mode versus Intelligence mode and review the token usage logs for each.
Who this matters for
- Vibe Builders: Set Auto mode to Intelligence for complex prompts, then check token usage logs to manage cost.
Harsh’s take
Static model selection is dead weight in code editors. Cursor Router fixes this by dynamically assigning tasks to cheaper or smarter models based on query complexity. Running every simple refactor through top-tier reasoning models drains budgets unnecessarily, while underpowered models fail complex architecture tasks.
This routing layer optimizes both speed and expenditure automatically. Engineering leads need to configure admin controls immediately to enforce Balance or Cost modes across team workspaces. Test the router's automatic picks against manual selections on complex codebases to ensure output quality remains high while cutting operational API costs.
by Harsh Desai
About Cursor
View the full Cursor page →All Cursor updatesGo deeper
More AI news
- FeatureHermes-agent adds tool self-recovery and raises iteration limits to 500
Hermes-agent enables tools to recover automatically from failures including truncated terminal outputs, whitespace mismatches in patches, and search near-misses. Default iteration limits increase from 90 to 500.
- Weekly DigestHermes Agent A2A protocol and voice barge-in, OpenClaw Wear OS app, and open agent tool fixes you can test now
Hermes Agent rolled out dozens of reliability, voice, and multi-agent features while OpenClaw added session rewinding, state recovery, and Wear OS support across the week.
- Daily RoundupBigBang-v1 trends on Hugging Face, Seedance 2.5 on Fal, and agent tools on Product Hunt
New image-text models and consistent video generators arrived alongside browser agents and storage tools, while a major chip investment signals hardware momentum for builders shipping AI products.