Back to the chart

incoai/splash

Local LLM inference engine for Apple silicon

Chart
Not on a chart right now
Stars
2.4k

Open on GitHubHomepage

What it is

Splash is a local inference engine for Apple silicon that runs coding agents and OpenAI- or Anthropic-compatible applications on a Mac. It combines DFlash 2 speculative decoding, specialized Metal kernels, and automatic memory planning.

Who it is for

Developers running coding agents or local LLM API servers on Apple M3 or newer Macs.

Why it is trending

Splash 1.7.0 was released on November 17, 2026, and the repository gained 51 stars today for a total of 2,404. It pairs each supported model family with a trained DFlash 2 draft and precompiled Metal kernels while automatically reusing cached prefixes and batching concurrent requests.

Main features

  • OpenAI and Anthropic compatible APIs with streaming, tool calls, and JSON Schema output
  • Vision support for images and PDFs, plus a built-in chat page and macOS desktop app
  • Supports GGUF (1-8 bit), MLX, and NVFP4 formats with automatic draft model selection
  • Automatic prefix cache reuse, concurrent request batching, and configurable memory and SSD caching

Summarized on November 19, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary