Can internlm3 8b instruct run on the RX 7900 XT?
Yes. At 8K context the best fit is Q8_0, using about 9.7 GB of the 19.0 GB usable on the RX 7900 XT. Estimated speed ceiling: ~85 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|
| 2-bit (Q2_K) • | 4.0 GB Fits | 4.2 GB Fits | 5.4 GB Fits | 232 |
| 3-bit (Q3_K_M) • | 4.9 GB Fits | 5.1 GB Fits | 6.2 GB Fits | 182 |
| 4-bit (Q4_K_M) • | 5.8 GB Fits | 6.0 GB Fits | 7.1 GB Fits | 149 |
| 5-bit (Q5_K_M) • | 6.7 GB Fits | 6.9 GB Fits | 8.0 GB Fits | 128 |
| 6-bit (Q6_K) • | 7.6 GB Fits | 7.8 GB Fits | 8.9 GB Fits | 111 |
| 8-bit (Q8_0) • | 9.5 GB Fits | 9.7 GB Fits | 10.9 GB Fits | 85 |
| 16-bit (F16/BF16) | 17.2 GB Tight | 17.4 GB Tight | 18.5 GB Tight | 45 |
• real GGUF file size from internlm/internlm3-8b-instruct-gguf. Speed is a bandwidth ceiling (800 GB/s), not a benchmark. Methodology.
More
internlm3 8b instruct 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 (internlm/internlm3-8b-instruct), AMD specs. Data as of 2026-11-17.