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.

MacBook Pro M1 Max 32GB wins for coding in balanced mode

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

SQwen 3.5 9B

llama.cppQ4_K_MTight fit

10.3 GB / 11.5 GB

8.0 tok/s25K ctx

AGemma 4 E4B

llama.cppQ4_K_MRuns well

8.8 GB / 11.5 GB

6.8 tok/s50K ctx

ACodeGeeX 4 9B

llama.cppQ4_K_MRuns well

8.7 GB / 11.5 GB

8.1 tok/s89K ctx

Winner

SDevstral Small 2 24B Instruct

llama.cppQ4_K_MTight fit

21.4 GB / 23.0 GB

16.2 tok/s27K ctx

SQwen 3.6 27B

llama.cppQ4_K_MTight fit

21.8 GB / 23.0 GB

11.0 tok/s36K ctx

SQwen 3.5 9B

llama.cppQ4_K_MRuns well

12.0 GB / 23.0 GB

43.1 tok/s96K ctx

Quick comparison