Local AI Coding app for Xcode

with AXIØM Vibe

  1. Launch the app

  2. Confirm one setting

  3. Copy/paste the port number into Xcode®

  4. Swift® Coding Assistance Chat appears right within Xcode® 🧑‍💻

30-Second Setup

Based on the Qwen series of coding models tuned for Swift Programming, AXIØM Vibe recommends a variant optimized for your Mac® hardware. Minimum is a M1 Air with 8gb of memory and the Recommended is a M1 Max with 64gb of memory or more.

macOS-Optimized LLMs

Can a local LLM compete with a frontier model? When it is optimized for Swift, SwiftUI®, and Apple® Frameworks the quality and speed of AXIØM Vibe will make API costs and usage-caps a thing of the past.

Swift AI Expert

“Each night I would hit that moment of USAGE EXCEEDED right as I got in the zone. Now, I can code as much or as little as I need, even offline, for less money. My business no longer feels chained.”

– Beta Tester

Features for VibeCoding compatible with Xcode

  1. Coding Intelligence is powered by Qwen 2.5 7B

    • Finetuned using QLoRA (Rank=16, Alpha=32) using a dataset of 5,602 high-quality, curated Swift examples

  2. Quantized into five variants to support the full range of Mac hardware with Apple Silicon:

    1. Compact at 3 bits for Macs with 8 gigabytes of memory

      1. all the way back to the MacBook Air or Mac mini with M1 chip

    2. Balanced at 4 bits for Macs with 16 or 18 gigabytes of memory

    3. Enhanced at 6 bits for Macs with 24 gigabytes of memory

    4. Studio at 8 bits for Macs with 32 or 36 gigabytes of memory

    5. Maximum at 16 bits for Macs with 64 gigabytes of memory or greater

  • Compatible with any Apple Silicon Mac running macOS 26.0+

  • AI coding assistance completely offline, local, and private

    • usage analytics are anonymous

  • Works right inside of Xcode via the officially-supported Local Chat Provider section of Xcode Settings > Intelligence

Or sign-up for notifications and get free access to the TestFlight® Beta! ↘︎