Qwen 3.8-27B takes 5 minutes to think on M5 MacBook Pro
talkaboutdesign · x · 2026-08-15
A user tested the Qwen3.8-27B model on an M5 MacBook Pro with 64GB of RAM. When asked a simple question about actors in programming, the model "thought" for 5 minutes without outputting a result, forcing the user to stop the process due to excessive friction. The user expressed anticipation for the day when such models can run fast on local hardware.
More from Infra
- AI agentic commerce requires both privacy and identity proof — provenauthority · 2026-08-15
- Anthropic's 5-Min Prompt Cache TTL Makes LLM Bills Up to 25% More Expensive — gethackteam · 2026-08-15
- Flock Cameras and Data Centers: When Useful Tech Loses the Narrative — DavidLinthicum · 2026-08-15
- User Laments: Can't Find Cheap Servers Anymore — oilmutt · 2026-08-15
- MiniMax H3 Benchmark: Native implementation 30%+ faster — Mattnix · 2026-08-15
- Intel explores HBM alternatives ZAM and XBM, production may take a decade — JOBhakdi · 2026-08-15