RTX 5090 vs M5 Max: best for local LLMs? #shorts

RTX 5090 (32 GB) vs MacBook Pro M5 Max (128 GB) — head-to-head for running large language models locally.

Each runs the model it's built for, at a similar quality tier. Token generation:
RTX 5090: Qwen 27B (dense) — 100 tok/s
M5 Max: Qwen 122B (A10B MoE) — 66 tok/s
Memory: RTX 5090 32GB vs M5 Max 128GB — the M5 Max fits far bigger models
Power draw (load): RTX 5090 575W vs M5 Max 90W — the M5 Max runs far cooler
Software stack: RTX 5090 battle-tested CUDA vs M5 Max younger MLX
Price: RTX 5090 $3,658 vs M5 Max $5,399 — the 5090 is cheaper

The 5090 wins on raw speed; the M5 Max fits far bigger models in its 128 gigs. It depends what you run.

So... which one would you buy? Drop a comment.

Full specs, benchmarks and the complete side-by-side: https://llmrequirements.com

#shorts #localllm #ai #llm #aihardware #gpu #machinelearning #selfhostedai #localai #rtx5090 #m5max #nvidia #apple

Видео RTX 5090 vs M5 Max: best for local LLMs? #shorts канала LLM Requirements