Products/Developer Tools / AI CLI/Show HN: Apfel – The free AI already on your Mac

Show HN: Apfel – The free AI already on your Mac

Free AI on Your Mac

Developer Tools / AI CLIOpen source / Self-funded100% On-Device (runs on Neural Engine + GPU)Zero cost - no API keys, no subscriptions, no per-token billingOpenAI-compatible HTTP server at localhost:11434CLI tool with stdin/stdout, JSON output, file attachmentsInteractive chat with context management and 5 trimming strategiesTool calling / function calling supportStreaming (SSE) support4,096 token context window (input + output combined)CORS for browser clientsDrop-in replacement for any OpenAI SDKBuilt-in demo scripts: cmd, oneliner, mac-narrator, explain, wtd, gitsum

Visit Show HN: Apfel – The free AI already on your Mac →

Our Take

This is genuinely clever — instead of waiting for Apple to ship proper AI dev tools, some dev just wrapped the hidden on-device LLM from macOS 26 and exposed it as a CLI and OpenAI-compatible API. The cool part: it runs entirely local on Apple Silicon's Neural Engine, zero cost, zero API keys, and you can plug it straight into any OpenAI SDK as a drop-in replacement. The 4,096 token context window is the trade-off, but for quick shell command generation, pipe chain building, or privacy-sensitive tasks where you don't want hitting cloud APIs, this is the move. The traction speaks for itself — engineers at Apple, Google, VMware, and NVIDIA are already using it.

A Swift binary that wraps Apple's FoundationModels framework to expose the built-in on-device LLM as a CLI tool, OpenAI-compatible HTTP server, and interactive chat interface

Problem It Solves

Apple ships an on-device LLM with macOS 26 (Tahoe) on Apple Silicon but only exposes it through Siri - apfel gives direct terminal and API access to it without needing to write Swift code

Target Customer

Developers, system administrators, and Mac power users with Apple Silicon Macs who want local AI inference without API costs

Use Cases

Natural language to shell command generation, Pipe chain generation from plain English, Command/error explanation, Directory/project orientation, Git commit summarization, OpenAI SDK integration, Scriptable AI workflows, Local privacy-sensitive AI tasks

Pricing Details

$0 token cost - uses Apple's built-in on-device LLM that ships with macOS

Free Tier

Yes - 100% free, uses local hardware

Differentiator

Leverages Apple's native on-device LLM that ships with macOS 26+ on Apple Silicon - no cloud, no API calls, no costs

Why Now

macOS 26 (Tahoe) is the first version to ship with Apple Intelligence including a built-in LLM accessible via FoundationModels framework - previously unavailable for direct developer access

Traction

Customers Mentioned: Engineers from Apple, Google, VMware, NVIDIA, and Grafana · Notable Metrics: 1.1k GitHub stars (1,030 stars as of April 3, 2026), 27 forks, 818 stars on April 3 alone

Key Facts

The people behind Show HN: Apfel – The free AI already on your Mac

Arthur-Ficial

profile

Links

Website GitHub Source: hacker-news

Want products like this in your inbox every morning?

Five products. Every morning. Written by someone who actually cares whether they're good or not. Free forever, unsubscribe whenever.