Show HN: Open-source engine running Gemma 4 26B in 2 GB RAM on any M-series Mac
Turbo Fieldfare is an open-source inference engine that enables running Google's Gemma 4 26B model on any M-series Mac with just 2 GB of RAM. The project, hosted on GitHub under the repository drumih/turbo-fieldfare, was shared on Hacker News as a Show HN post. By optimizing memory usage, it allows the 26-billion-parameter model to operate within the constraints of Apple's unified memory architecture, which is typically limited on base-model Macs. This makes it possible to run a large language model locally without requiring a dedicated GPU or cloud-based inference. The engine is designed to be efficient and accessible, potentially lowering the barrier for developers and researchers to experiment with Gemma 4 on personal laptops. The source text does not provide details on performance benchmarks, inference speed, or specific optimizations used, but the achievement highlights a significant reduction in hardware requirements for running large models.
Enables running a 26B parameter model locally on consumer Macs with minimal RAM.