Can Qwen3.5 9B run on the RX 7900 XT?
Yes. At 8K context the best fit is Q8_0, using about 10.0 GB of the 19.0 GB usable on the RX 7900 XT. Estimated speed ceiling: ~84 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) | 4.4 GB Fits | 4.5 GB Fits | 5.2 GB Fits | 8.2 GB Fits | 221 |
| 3-bit (Q3_K_M) • | 5.4 GB Fits | 5.5 GB Fits | 6.2 GB Fits | 9.2 GB Fits | 171 |
| 4-bit (Q4_K_M) • | 6.3 GB Fits | 6.4 GB Fits | 7.2 GB Fits | 10.2 GB Fits | 141 |
| 5-bit (Q5_K_M) • | 7.1 GB Fits | 7.2 GB Fits | 8.0 GB Fits | 11.0 GB Fits | 122 |
| 6-bit (Q6_K) • | 7.9 GB Fits | 8.1 GB Fits | 8.8 GB Fits | 11.8 GB Fits | 107 |
| 8-bit (Q8_0) • | 9.9 GB Fits | 10.0 GB Fits | 10.7 GB Fits | 13.7 GB Fits | 84 |
| 16-bit (F16/BF16) • | 17.7 GB Tight | 17.8 GB Tight | 18.6 GB Tight | 21.6 GB No | 45 |
• real GGUF file size from unsloth/Qwen3.5-9B-GGUF. Speed is a bandwidth ceiling (800 GB/s), not a benchmark. Methodology.
More
Qwen3.5 9B on every GPU and Mac · Every model that runs on the RX 7900 XT · Other models that fit 24 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (Qwen/Qwen3.5-9B), AMD specs. Data as of 2026-11-17.