Replit makes intelligent model routing its default, automatically choosing the best AI for each coding task

Replit is rolling out a new system-wide default that automatically determines which underlying AI model should handle each coding request. The feature, called Auto mode, makes routing decisions on the fly by weighing performance quality, execution speed, and operational expense.

While the company is pushing Auto mode as the standard for all users, paid tiers—Core and Pro subscribers—retain the ability to override the system and manually select specific models when they need direct control over which AI powers their work.

Why routing is taking off#

The move reflects a broader wave of investment in model-routing systems across the AI industry. Within a single week in August, both Stripe announced its acquisition of OpenRouter for $8 billion and Ramp unveiled Router.com, a service designed to direct requests to the lowest-cost models that still meet performance thresholds. SpaceX-owned Cursor rolled out its own router earlier in the summer, while reports suggest Meta is developing an internal routing system dubbed "Switchboard."

The economic calculus driving this trend is straightforward: pricing differences within a model family can span orders of magnitude, yet smaller, cheaper alternatives now approach the capability level of larger models. Michele Catasta, Replit's president and head of AI, explains the opportunity: "Across one model family, per-token rates can span orders of magnitude. At the same time, the intelligence of cheaper, smaller models is now much closer to their larger frontier counterparts, providing us a lot of room for cost optimizations."

How Auto mode works#

Replit rolled out the foundation for this shift when it launched Free Mode last week—a lower-cost, Agent-based option that automatically selects models without consuming subscription credits, though subject to usage caps that refresh every five hours. Free Mode is now becoming standard across all user accounts.

Paid subscribers (Core and Pro tiers) can upgrade to Power or Max modes to disable Auto mode and manually pick models, potentially triggering additional costs. Replit's system can also recommend stepping up to higher-tier modes when it judges that a task demands more computational power. For Enterprise customers, workspace administrators can restrict routing to an approved subset of models, maintaining both automation and policy compliance.

A key advantage of Replit's approach: . Rather than making a single model choice at the start, the system can swap between different models mid-task to optimize the trade-off between capability and expense at different stages.