Compare

Compare local AI hardware with workload-aware output.

Operating mode: Balanced. Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Mac mini M2 24GB wins for coding in balanced mode

Based on model fit, speed, and quality across top recommendations.

Winner

SQwen 3.5 9B

llama.cppQ4_K_MRuns well

11.2 GB / 17.3 GB

12.7 tok/s60K ctx

ANemotron Nano 9B v2

llama.cppQ4_K_MRuns well

11.4 GB / 17.3 GB

12.7 tok/s54K ctx

AGemma 4 E4B

llama.cppQ4_K_MRuns well

9.7 GB / 17.3 GB

10.9 tok/s111K ctx

SQwen 3.5 9B

llama.cppQ4_K_MTight fit

10.3 GB / 11.5 GB

12.7 tok/s25K ctx

AGemma 4 E4B

llama.cppQ4_K_MRuns well

8.8 GB / 11.5 GB

10.9 tok/s50K ctx

ACodeGeeX 4 9B

llama.cppQ4_K_MRuns well

8.7 GB / 11.5 GB

12.9 tok/s89K ctx

Quick comparison