Lindy switched from Anthropic to DeepSeek and saved millions

By June 2026, U.S. developers and startups were looking for cheaper AI models because common tasks did not always justify premium-model prices. Lindy's founder Flo Crivello said the company had been using Anthropic models.

Lindy switched to DeepSeek. Crivello said the move saved the firm millions of dollars and argued that lower tiers of intelligence were sufficient for routine work at roughly a tenth of the price.

The shift turned model selection into a cost-control lever. Lindy accessed DeepSeek through an American provider after privacy questions, pairing the savings decision with a data-routing safeguard.

Lindy's founder said moving from Anthropic models to DeepSeek saved the firm millions of dollars.

Many coding and assistant tasks could be handled by cheaper models even when frontier models remained better at complex work.

The price gap made premium-model usage hard to justify for routine tasks.

DeepSeek's low cost made it practical for a startup to test a different model supplier.

Using an American provider addressed questions about whether user data was sent directly to China.

Treat model choice as a portfolio decision: use a cheaper model when quality clears the task threshold, and reserve premium capacity for work that truly needs it.

Rest of World reported that Crivello attributed millions of dollars in savings to the switch, but it did not provide a later revenue or margin figure. Lindy's use of an American provider showed how the company could pair the model migration with a data-routing safeguard.