deepseek-v4.1POST

Cost math: local DeepSeek V4.1 Flash on a 256GB M5 Ultra vs API pricing

The author estimates ~35 tok/s and ~1.1B tokens/year locally on a 256GB M5 Ultra, costing $394 in electricity versus ~$663/year for equivalent off-peak API output, concluding the laptop is overspend if bought only for local LLMs.

George Kiknadze

@kiknaioArchived 6 days agoOriginally posted 2026.10.24

George Kiknadze @kiknaioOriginally posted 2026.10.24

Open original on X

Local DeepSeek V4.1 Flash on that 256GB M5 Ultra (rough math): ~35 tok/s decode (full model doesn’t fit) Running it 24/7: ~1.10 billion tokens/year API equivalent (official off-peak output @ $0.60/M): ~$663/year Macbook electricity consumption (250W, ~$0.18/kWh): $394/year. From model running perspective I really don't think it's feasible. I think it's an amazing laptop and you really would enjoy using it and you'll be super productive but purchasing that only for local LLM seems like overspending to me.

x.com/kiknaio/status/2113986004012995050Archived 2026.10.25 12:02