Swiftlet is listed as Open Source. Check https://github.com/leonickson1/Swiftlet for the current plan limits, as vendors change pricing frequently.

Swiftlet

Q: What is Swiftlet?

Swiftlet is a code AI tool listed on Use AI Tools. Swiftlet is a Swift and Metal runtime that runs large Qwen mixture-of-experts models locally on Apple devices by streaming expert weights from storage. Because experts are read from disk instead of held in memory, RAM use stays low enough to run 35B and 80B models, including on iPhone.

Q: What is Swiftlet best for?

Swiftlet is commonly used for Apple device developers, Large mixture-of-experts models, Inference without the cloud. It supports English.

Code⚡ Advanced✅ Global access

Apple device developersLarge mixture-of-experts modelsInference without the cloud

Open SourceEnglish

Swiftlet is a Swift and Metal runtime that runs large Qwen mixture-of-experts models locally on Apple devices by streaming expert weights from storage. Because experts are read from disk instead of held in memory, RAM use stays low enough to run 35B and 80B models, including on iPhone.

Visit Website Compare

Community Rating

Be the first to rate Swiftlet. Your feedback helps other users decide.

🎯 Real-World Use Cases

Run an 80B Model on a Mac

Stream expert weights instead of holding them in RAM

Prototype on iPhone

Test large models on mobile Apple hardware

Avoid Cloud Costs

Keep inference on hardware you already own

⚖️ Honest Review

✅ Pros

+Runs models larger than the RAM you have
+Native Swift and Metal on Apple Silicon
+Works on iPhone as well as Mac

⚠️ Cons

−Tuned around Qwen mixture-of-experts models
−Large weights must be installed on local storage

📚

Guides and comparisons about code tools

Local inference running on consumer hardware

CodeOct 18, 2026