Swiftlet is an open-source project whose Show HN title claims that an 80B-parameter Qwen model can run in 4.3 GB of RAM on a Mac, while a 35B model can run on an iPhone. The supplied Hacker News snapshot reports 304 points and 137 comments. However, the available metadata does not explain the compression or inference method, token generation speed, quality loss, device configuration, or reproducibility procedure, so the headline results should be treated as project claims pending independent verification.
No heat snapshots are available in the last 24 hours.