Can V100 32GB SXM2 run Llama-3.1-70B-Instruct?

NVIDIA · 32 GB HBM2 · 900 GB/s bandwidth · Meta · 70.6B · model ctx up to 128K

Fit & speed by quant and context

Fits = weights + KV(ctx) + 1 GB overhead ≤ 95% of VRAM. Measured anchors are context-agnostic; the fit verdict is recomputed per context. A missing tok/s means the model is far beyond this card (offload-only territory).

VRAM breakdown at 8K context

Other GPUs that run Llama-3.1-70B-Instruct

• B200 192GB SXM — ≈520.0 tok/s Q4 measured
• MI300X 192GB — ≈77.7 tok/s Q4 est.
• H200 141GB SXM5 — ≈74.4 tok/s Q4 est.
• H100 80GB SXM5 — ≈51.9 tok/s Q4 est.
• Gaudi 3 128GB — ≈48.0 tok/s Q4 est.
• Mac Studio M3 Ultra 96GB — ≈14.1 tok/s Q4 measured
• MacBook Pro M4 Max 128GB — ≈12.0 tok/s Q4 measured
• MacBook Pro M3 Max 128GB — ≈9.0 tok/s Q4 measured

All GPUs for Llama-3.1-70B-Instruct →

Other models for the V100 32GB SXM2

• MiniCPM5-1B — ≈895.6 tok/s Q4 est.
• Qwen3.5-2B — ≈483.6 tok/s Q4 est.
• DeepSeek-R1-0528-Qwen3-8B-layer-mix-bpw-3.8-mlx — ≈420.5 tok/s Q4 est.
• Qwen3.6-35B-A3B — ≈322.4 tok/s Q4 est.
• Qwen3.5-35B-A3B — ≈322.4 tok/s Q4 est.
• GLM-4.7-Flash — ≈322.4 tok/s Q4 est.
• Qwen3-30B-A3B-Instruct-2507 — ≈322.4 tok/s Q4 est.
• Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 — ≈322.4 tok/s Q4 est.

All models on the V100 32GB SXM2 →