Can diffusiongemma 26B A4B it run on the RTX 4070 Ti Super?
Yes. At 8K context the best fit is Q3_K_M, using about 13.2 GB of the 15.2 GB usable on the RTX 4070 Ti Super. Estimated speed ceiling: ~344 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) | 10.3 GB Fits | 10.4 GB Fits | 11.4 GB Fits | 15.1 GB Tight | 448 |
| 3-bit (Q3_K_M) | 13.0 GB Fits | 13.2 GB Fits | 14.1 GB Tight | 17.8 GB No | 344 |
| 4-bit (Q4_K_M) • | 16.9 GB No | 17.1 GB No | 18.0 GB No | 21.7 GB No | — |
| 5-bit (Q5_K_M) • | 19.1 GB No | 19.2 GB No | 20.2 GB No | 23.9 GB No | — |
| 6-bit (Q6_K) • | 22.4 GB No | 22.5 GB No | 23.4 GB No | 27.2 GB No | — |
| 8-bit (Q8_0) • | 26.3 GB No | 26.4 GB No | 27.4 GB No | 31.1 GB No | — |
| 16-bit (F16/BF16) • | 48.3 GB No | 48.5 GB No | 49.4 GB No | 53.2 GB No | — |
• real GGUF file size from unsloth/diffusiongemma-26B-A4B-it-GGUF. Speed is a bandwidth ceiling (672 GB/s), not a benchmark. Methodology.
More
diffusiongemma 26B A4B it on every GPU and Mac · Every model that runs on the RTX 4070 Ti Super · Other models that fit 16 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (google/diffusiongemma-26B-A4B-it), NVIDIA specs. Data as of 2026-11-17.