My daily average local model token burn is 17M They have become tremendously useful.

Clem Delangue reports consuming approximately 17 million tokens daily when running local language models, indicating substantial usage of on-device AI inference. He expresses satisfaction with the utility and practical value these local models provide for his workflows. This suggests growing viability and effectiveness of running inference locally rather than relying solely on cloud-based AI services.

Source: Clem Delangue (X) | 2026-05-24