GPU

Apple MacBook Pro M4 Max (40-core GPU) 48GB for local AI

36GB memory, 546 GB/s bandwidth, and deterministic compatibility links for downloadable local models.

VRAM

36GB

Bandwidth

546 GB/s

Architecture

Apple M4 Max

Memory type

LPDDR5X Unified

Models to check on this GPU

• Can it run OpenAI gpt-oss-20b?
• Can it run Kimi-Linear-48B-A3B-Instruct?
• Can it run Phi-3.5-MoE-instruct?
• Can it run Hermes-4.3-36B?
• Can it run Intern-S2-Preview?
• Can it run Qwen3.5-35B-A3B?
• Can it run Qwen3.6-35B-A3B?
• Can it run Qwen-AgentWorld-35B-A3B?
• Can it run Nous-Capybara-34B?