llamaperf

NVIDIA B200 180GB reports

1 report. These are individual setups, not a controlled benchmark. NVIDIA B200 180GB speeds and models →

Speeds are as each source reported them. GPU count, offloading and concurrent requests change them, and prompt-processing speed is input throughput, not time to first token. Check the full setup before comparing two reports.

Tone: positiveNVIDIA hardware
generation:
4.8 tokens/s
quant:
NVFP4
vision

User reports training a 40.1M connector to give DeepSeek V4 Flash basic vision, with the model loaded across four B200s in a custom SGLang stack. Training ran on 5x H200. User also trained a Laguna XS 2.1 version (33B total / 3B active) with a 30.7M connector, inspired by Baseten's GLM-5.2 Vision NVFP4.

Sep 7, 2026