Local versus cloud

Local AI vs cloud API: when to use which route

Use local inference when the model fits cleanly and privacy, latency, or no-per-token cost matters. Use cloud/API fallback when hardware is red, context is long, or reliability matters more than local control.

Open compatibility calculator

24GB clean fits27
Commercial rankingfalse
Cloud cost companionapiroute.dev
Red examples on 12GBgpt-oss-safeguard 120B, Devstral 2 123B, GLM-5.2
Green

Prefer local inference first when there is clean headroom.

Yellow

Reduce context, lower quantization, or expect offload slowdown.

Red

Use a smaller local model, larger hardware, or cloud/API fallback.

Fallback policy

Commercial options are disclosed follow-ups, never ranking input.
3 options