GPU × use-case guide · Vision-language

Is the B200 NVL (pair) a good GPU for vision-language?

The B200 NVL (pair) is a blackwell NVIDIA GPU with 360 GB HBM3e (16000 GB/s), 4500 TFLOPS BF16 / 9000 TFLOPS FP8, and a 1800 W TDP. Vision-language workloads care most about extra VRAM for the vision encoder and image tokens on top of the language model. Here's how the B200 NVL (pair) measures up.

What models fit on a single B200 NVL (pair)?

Weights only, reserving ~25% of the 360 GB for KV cache, activations and fragmentation. ✓ = fits on one card.

Largest single-card fit: Qwen 2.5 72B at BF16, Qwen 2.5 72B at FP8, Llama 3.1 405B at INT4. Bigger models need tensor-parallel across 4 cards.

B200 NVL (pair) for vision-language, specifically

Vision-language is context-heavy, so the KV cache — not the weights — is what fills the 360 GB. On the B200 NVL (pair) you'll trade context length against batch size: long prompts mean fewer concurrent requests. Because it runs offline, batch aggressively to push tokens-per-dollar down. Size it precisely on the calculator.

B200 NVL (pair) pricing across providers

Verdict

With 360 GB, the B200 NVL (pair) is a data-center-class card that comfortably handles vision-language for models up to Qwen 2.5 72B at full precision on a single card — a strong pick if your budget supports ~$11.49/hr.