modelbenchmark.io

Nemotron 3.5 Super VL

NVIDIA · nvidia-nemotron-3-5-super-vl-120b-a12b

Compare

Nemotron 3.5 Super VL is a multimodal MoE model built for visual understanding, reasoning, and agentic workflows.

Specification

most-agreed values

Context
1M
Max output
1M
Released
2026-10-15
Knowledge cutoff
—
Retires
—
Open weights
yes
Input
text, image
Output
text

Price

US dollars per million tokens · most-agreed

Input
$0.15
Output
$0.50
Cache read
$0.02

Available from 4 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
wandb$0.15$0.50$0.02—1M1M—
baseten$0.20$0.50$0.04—1M1M—
venice$0.20$0.50$0.04—262K33K—
wandb · wandb$0.25$1.50$0.05—1M944K—

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-10-16wandb · wandbCache read$0.02
2026-10-29wandb · wandbCache read$0.05
2026-10-16wandb · wandbInput$0.15
2026-10-29wandb · wandbInput$0.25
2026-10-16wandbOutput$0.75
2026-10-17wandbOutput$0.50
2026-10-16wandb · wandbOutput$0.75
2026-10-29wandb · wandbOutput$1.50
2026-10-17veniceCache read$0.125
2026-10-28veniceCache read$0.04
2026-10-17veniceInput$0.25
2026-10-28veniceInput$0.20
2026-10-17veniceOutput$1.25
2026-10-28veniceOutput$0.50

More from NVIDIA

most-hosted first