The PCPartPicker for Local AI
Can your PC run
this AI model?
Enter your hardware and instantly see which local models fit — with performance estimates and one-line install commands.
The PCPartPicker for Local AI
Can your PC run
this AI model?
Enter your hardware and instantly see which local models fit — with performance estimates and one-line install commands.
►
llama3.1:8b
excellent · ~62 tok/s
qwen2.5-coder:14b
great · ~38 tok/s
mistral-nemo:12b
good · ~29 tok/s
gemma3:27b
excellent · ~21 tok/s
deepseek-r1:7b
great · ~55 tok/s
phi4:14b
good · ~33 tok/s
llava:13b
great · ~30 tok/s
gemma3:12b
excellent · ~41 tok/s
mistral:7b
great · ~58 tok/s
codellama:13b
good · ~35 tok/s
llama3.1:8b
excellent · ~62 tok/s
qwen2.5-coder:14b
great · ~38 tok/s
mistral-nemo:12b
good · ~29 tok/s
gemma3:27b
excellent · ~21 tok/s
deepseek-r1:7b
great · ~55 tok/s
phi4:14b
good · ~33 tok/s
llava:13b
great · ~30 tok/s
gemma3:12b
excellent · ~41 tok/s
mistral:7b
great · ~58 tok/s
codellama:13b
good · ~35 tok/s
llama3.1:8b
excellent · ~62 tok/s
qwen2.5-coder:14b
great · ~38 tok/s
mistral-nemo:12b
good · ~29 tok/s
gemma3:27b
excellent · ~21 tok/s
deepseek-r1:7b
great · ~55 tok/s
phi4:14b
good · ~33 tok/s
llava:13b
great · ~30 tok/s
gemma3:12b
excellent · ~41 tok/s
mistral:7b
great · ~58 tok/s
codellama:13b
good · ~35 tok/s
onyxmatch — resultRTX 4070 · 12GB
$ onyx scan
✓llama3.1:8b
excellent · ~62 tok/s
✓qwen2.5-coder:14b
great · ~38 tok/s
✓mistral-nemo:12b
good · ~29 tok/s
·llama3.1:70b
needs 40GB+ VRAM
$ ollama run llama3.1:8b