Can Gemma 3 4B IT run on the H100 SXM?
Yes. At 8K context the best fit is F16, using about 8.4 GB of the 76.0 GB usable on the H100 SXM. Estimated speed ceiling: ~431 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) • | 2.7 GB Fits | 2.8 GB Fits | 3.2 GB Fits | 5.1 GB Fits | 1937 |
| 3-bit (Q3_K_M) • | 3.0 GB Fits | 3.1 GB Fits | 3.6 GB Fits | 5.5 GB Fits | 1596 |
| 4-bit (Q4_K_M) • | 3.4 GB Fits | 3.5 GB Fits | 4.0 GB Fits | 5.8 GB Fits | 1345 |
| 5-bit (Q5_K_M) • | 3.7 GB Fits | 3.8 GB Fits | 4.3 GB Fits | 6.1 GB Fits | 1184 |
| 6-bit (Q6_K) • | 4.1 GB Fits | 4.1 GB Fits | 4.6 GB Fits | 6.5 GB Fits | 1050 |
| 8-bit (Q8_0) • | 4.9 GB Fits | 5.0 GB Fits | 5.5 GB Fits | 7.4 GB Fits | 811 |
| 16-bit (F16/BF16) • | 8.3 GB Fits | 8.4 GB Fits | 8.9 GB Fits | 10.7 GB Fits | 431 |
• real GGUF file size from unsloth/gemma-3-4b-it-GGUF. Speed is a bandwidth ceiling (3350 GB/s), not a benchmark. Methodology.
What fits instead
- Cheapest listed GPU that runs Gemma 3 4B IT: Arc B580 (12 GB, launch MSRP $249) at F16, fully in VRAM at 8K context.
- Largest popular model that fits the H100 SXM: gpt-oss 120B (117B, MoE) at F16, about 61.9 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
Gemma 3 4B IT on every GPU and Mac · Every model that runs on the H100 SXM · Other models that fit 64 GB VRAM · Try other settings in the calculator
Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs
Data: Hugging Face Hub (google/gemma-3-4b-it), NVIDIA specs. Model data last checked 2026-11-17; refreshed hourly.