local.aiEarly access

VIP

beshkenadze

local.ai/beshkenadze

Claimed August 2026

My machines

MacBook Pro M1 Max
Approximate
Apple · 32 GB
Approximate options
Choose what matters most
Best fit
Gemma 4 31B It · AWQ-4BIT
Best balance of speed and intelligence.
Intelligence53.8
Speed70.5 tasks/hr
Size20.9 GB
More intelligent
Qwen3.6 35B A3B · UD-IQ4_XS
~27% more intelligence · ~22% slower.
Intelligence68.3
Speed58.0 tasks/hr
Size17.7 GB
Faster
Holo 3.1 35B A3B · NVFP4
~15% faster · retains ~99% intelligence.
Intelligence53.0
Speed83.1 tasks/hr
Size23.7 GB
Approximate for 32 GB VRAM. Results may vary.
NVIDIA GeForce RTX 4090 24GB
NVIDIA · 24 GB
Recommended for NVIDIA GeForce RTX 4090 24GB
Choose what matters most
Best fit
Qwen3.5 9B · Q4_K_M
Best balance of speed and intelligence.
Intelligence50.5
Speed41.5 tasks/hr
Size5.7 GB
More intelligent
Gemma 4 26B A4B It · AWQ-4BIT
6% more intelligence · 1% slower.
Intelligence53.5
Speed41.1 tasks/hr
Size17.2 GB
Faster
Gemma 4 E4B It · Q4_K_M
77% faster · retains 57% intelligence.
Intelligence28.9
Speed179 tasks/hr
Size5.0 GB
Compare all models