local.aiEarly access

localinference

local.ai/localinference

Claimed August 2026

My machines

16" M5 Max Macbook Pro
Apple · 128 GB
Recommended for MacBook Pro M5 Max 128GB · 40-core GPU
Choose what matters most
Best fit
Qwen3.6 35B A3B · UD-Q4_K_S
Best balance of speed and intelligence.
Intelligence68.5
Speed17.0 tasks/hr
Size20.9 GB
More intelligent
Step 3.7 Flash · IQ4_XS
2% more intelligence · 31% slower.
Intelligence70.0
Speed13.0 tasks/hr
Size105.0 GB
Faster
Ornith 1.0 35B · Q4_K_M
37% faster · retains 98% intelligence.
Intelligence67.2
Speed27.1 tasks/hr
Size21.2 GB
Compare all models
DGX Spark
×3
Nvidia · 128 GB each
Recommended for NVIDIA DGX Spark 128GB
Choose what matters most
Best fit
Qwen3.6 35B A3B · NVFP4
Best balance of speed and intelligence.
Intelligence69.7
Speed21.8 tasks/hr
Size23.4 GB
More intelligent
DeepSeek V4 Flash 0731 · UD-IQ2_M
12% more intelligence · 280% slower.
Intelligence77.8
Speed5.7 tasks/hr
Size90.9 GB
Faster
NVIDIA Nemotron 3.5 Lightning 30B A3B · NVFP4
29% faster · retains 81% intelligence.
Intelligence56.2
Speed30.6 tasks/hr
Size21.6 GB
Compare all models