TurboFieldfare is a Swift and Metal runtime that runs Gemma 4 26B-A4B in about 2 GB of RAM on any Apple Silicon Mac, including the 8 GB models. It is built for memory-constrained local inference, ships a Mac app and a local OpenAI-compatible server, and documents its system design, benchmarks and optimization work in the repository.
Be the first to rate TurboFieldfare. Your feedback helps other users decide.
Load Gemma 4 26B-A4B on a base M-series MacBook
Expose an OpenAI-compatible endpoint from your Mac
Build Swift apps that call a model running on the same machine
Similar tools in the same category — compare side-by-side
PS5LM ports llama.cpp to a jailbroken PlayStation 5 so the console GPU runs GGUF models and serves t...
Swiftlet is a Swift and Metal runtime that runs large Qwen mixture-of-experts models locally on Appl...
Janas is a pure C inference runtime for running language models larger than your RAM, on CPU or GPU....
Guides and comparisons about code tools
Eight local inference projects that fit big models into small memory: a 26B model in 2 GB, MoE weight streaming, a console as a server and sub-1-bit compression
Discover the best AI code tools for API development in 2026. Compare tools for API design, documentation, testing, monitoring, and security automation.
Discover 5 AI tools for deployment automation and CI/CD in 2026. Streamline cloud deployments with zero-downtime strategies and rollback management.
Discover 5 AI tools for cloud security in 2026. Detect misconfigurations, scan vulnerabilities, and ensure compliance across AWS, GCP, and Azure.
Discover 5 AI tools for data science in 2026. Automate EDA, feature engineering, model selection, and notebook workflows with AI-powered assistance.
