Can granite 4.0 h small run on the 2x RTX 4090?
Yes. At 8K context the best fit is Q8_0, using about 32.6 GB of the 45.6 GB usable on the 2x RTX 4090. Estimated speed ceiling: ~108 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) • | 11.6 GB Fits | 11.7 GB Fits | 12.1 GB Fits | 13.6 GB Fits | 313 |
| 3-bit (Q3_K_M) • | 15.0 GB Fits | 15.0 GB Fits | 15.4 GB Fits | 16.9 GB Fits | 240 |
| 4-bit (Q4_K_M) • | 18.8 GB Fits | 18.9 GB Fits | 19.2 GB Fits | 20.7 GB Fits | 189 |
| 5-bit (Q5_K_M) • | 21.9 GB Fits | 22.0 GB Fits | 22.4 GB Fits | 23.9 GB Fits | 161 |
| 6-bit (Q6_K) • | 25.3 GB Fits | 25.4 GB Fits | 25.7 GB Fits | 27.2 GB Fits | 139 |
| 8-bit (Q8_0) • | 32.6 GB Fits | 32.6 GB Fits | 33.0 GB Fits | 34.5 GB Fits | 108 |
| 16-bit (F16/BF16) • | 60.7 GB No | 60.7 GB No | 61.1 GB No | 62.6 GB No | — |
• real GGUF file size from ibm-granite/granite-4.0-h-small-GGUF. Speed is a bandwidth ceiling (1008 GB/s), not a benchmark. Methodology.
More
granite 4.0 h small on every GPU and Mac · Every model that runs on the 2x RTX 4090 · Try other settings in the calculator
Data: Hugging Face Hub (ibm-granite/granite-4.0-h-small), NVIDIA specs. Data as of 2026-11-17.