Rapid-MLXFrom the builder
How Fast Are Qwen3.8 27B, Qwen3.6 35B and DeepSeek V4.1 on a Mac? 4 Macs Tested
Rapid-MLX ran the same 1.0 build and benchmark across four Macs—an M2 Pro, M4 Pro, M5 Max, and M3 Ultra—to compare generation speed for popular local models, with every number linked to a public run on the community leaderboard. The post reports tokens-per-second for models such as LFM3, Qwen3.8 27B, Qwen3.6 35B, and DeepSeek V4.1 on each machine.
Chip generation and unified memory strongly shape throughput, and the same model can differ by several times across the lineup. The authors note that speed is not quality, and point readers to separate posts on coding quality and RAM-based model picks.
We ran the Rapid-MLX 1.0 benchmark on four Macs: an M2 Pro Mac mini (32 GB), an M4 Pro Mac mini (48 GB), an M5 Max Mac Studio (36 GB) and an M3 Ultra Mac Studio (256 GB).
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.