Gemma 4 31B MacBook test

작성: Estebankiwi: GPT

Gemma 4 31B proves more intelligent than GPT 5 was. And I am running it locally on a MacBook. I just tested Gemma 4 31B and Qwen 3.6 35B on my $5,399 MacBook Pro M5 Max with 128GB unified memory. The speeds are insane. Over 100 tokens per second. But here is the catch. Benchmark intelligence and local performance are two different things. The models test well on paper but run worse locally than they do in a data center. Local AI is 9 months behind frontier. That gap is shrinking fast. But for production vibe coding, GPT 5.5 and Claude Opus 4.7 are still the move. Full review and real workflow test below.