Cost Analysis: Running Local LLMs on Apple Silicon vs. OpenRouter API
By
datadrivenangel
14d ago· 3 min readenInsight
65/100
Toasty
Bagelometer↗
A good honest bake. Not flashy, but you'll finish the whole bagel.
Score65TypeanalysisSentimentneutral
Summary
The article compares the cost of running local LLMs on Apple Silicon (specifically an M5 MacBook Pro) versus using OpenRouter's API. The author calculates that running local models costs approximately $1.50 per million tokens when factoring in electricity and accelerated device depreciation, while OpenRouter offers comparable models at 1/3 the price with roughly 2x the speed. The article is part 3 of a series on "Offline Agentic Coding" and argues that despite the appeal of local LLMs, cloud-based solutions like OpenRouter are more cost-effective for most use cases.
Key quotes
· 4 pulledApple silicon costs more than OpenRouter.
At a few tens of tokens per second this works out to ammortized costs of ~$1.50 per million tokens.
Openrouter for comparable models is 1/3rd the price and ~2x the speed.
Accelerated depreciation (if any) from shortening the lifespan of the device will be more expensive than the electricity.
Local LLMs can be very very cheap
