LOCAL AIregistry
‹ all GPUs

RTX 4080 SUPER

NVIDIA · 16 GB · 736 GB/s
Top 2 recipes
Recommended
Gemma 4 26B A4B
EXL3 3.1 bpwTabbyAPI32K context

Best for Fast multimodal work on 16–24 GB cards.

100 tok/sdecode
1,995 tok/sprefill
32Kcontext
✓ Thinks✓ Tools✓ Long context✓ Vision
Tested on this card · Sep 26, 2026
Longest context
Qwen3.8-27B
EXL3 3 bpwTabbyAPI128K context

Best for Coding agents, long documents, everyday assistant work.

77 tok/sdecode
–prefill
128Kcontext
ThinksToolsLong context✓ Vision
Earlier check · Sep 22, 2026