Can gemma 3 12b it run on the RX 7900 XTX?
Yes. At 8K context the best fit is Q8_0, using about 13.4 GB of the 22.8 GB usable on the RX 7900 XTX. Estimated speed ceiling: ~77 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) • | 5.9 GB Fits | 6.2 GB Fits | 7.7 GB Fits | 13.7 GB Fits | 201 |
| 3-bit (Q3_K_M) • | 7.1 GB Fits | 7.3 GB Fits | 8.8 GB Fits | 14.8 GB Fits | 160 |
| 4-bit (Q4_K_M) • | 8.3 GB Fits | 8.5 GB Fits | 10.0 GB Fits | 16.0 GB Fits | 131 |
| 5-bit (Q5_K_M) • | 9.3 GB Fits | 9.6 GB Fits | 11.1 GB Fits | 17.1 GB Fits | 114 |
| 6-bit (Q6_K) • | 10.5 GB Fits | 10.7 GB Fits | 12.2 GB Fits | 18.2 GB Fits | 99 |
| 8-bit (Q8_0) • | 13.1 GB Fits | 13.4 GB Fits | 14.9 GB Fits | 20.9 GB Tight | 77 |
| 16-bit (F16/BF16) • | 23.4 GB No | 23.6 GB No | 25.1 GB No | 31.1 GB No | — |
• real GGUF file size from unsloth/gemma-3-12b-it-GGUF. Speed is a bandwidth ceiling (960 GB/s), not a benchmark. Methodology.
More
gemma 3 12b it on every GPU and Mac · Every model that runs on the RX 7900 XTX · Other models that fit 24 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (google/gemma-3-12b-it), AMD specs. Data as of 2026-11-17.