Custom software links iPhone to MacBook for faster AI model runs
A developer used custom open-source software, 'backburner,' to split an AI workload between an iPhone 17 Pro Max and a MacBook Pro with Apple's M4 Pro chip. By offloading some AI tasks to the phone, they achieved up to a 44% jump in prefill speed for large language models. The iPhone's chips, including its Neural Engine, handled later model layers more efficiently than the Mac alone. Main caveat: it's only faster for initial prefill, not ongoing text generation.
- iPhone 17 Pro Max connected to MacBook via USB-C
- Open-source tool 'backburner' splits AI tasks
- Prefill speed gains peaked at 44%
- Only benefits initial model loading, not text output
- More powerful iPhones could further improve results
Sources covering this
In this story
More in AI
Anthropic adds always-on agent abilities to Claude
Anthropic has quietly added always-on AI agent features to its Claude assistant, letting it handle ongoing or scheduled tasks even after…
Pope Leo XIV rejects AI-generated art, urges support for human artists
Pope Leo XIV publicly criticized AI-generated art, calling for a clear distinction between work made by humans and images produced by…
Anthropic commits $100M to train 10,000 AI engineers by 2027
Anthropic’s putting $100 million into Claude Frontier Academy, a new program meant to train 10,000 engineers from partner companies—like…
Google unveils Gemini 4 Argon AI for cybersecurity
Google introduced Gemini 4 Argon, its new frontier AI model with a 1-million-token output limit and advanced reasoning, focusing first…