Can DeepSeek V2 Lite Chat run on the RTX 5090?
Yes. At 8K context the best fit is Q8_0, using about 16.4 GB of the 30.4 GB usable on the RTX 5090. Estimated speed ceiling: ~633 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) • | 6.7 GB Fits | 6.8 GB Fits | 7.5 GB Fits | 10.4 GB Fits | 1645 |
| 3-bit (Q3_K_M) • | 8.3 GB Fits | 8.4 GB Fits | 9.1 GB Fits | 12.0 GB Fits | 1301 |
| 4-bit (Q4_K_M) • | 10.4 GB Fits | 10.5 GB Fits | 11.2 GB Fits | 14.0 GB Fits | 1020 |
| 5-bit (Q5_K_M) • | 11.8 GB Fits | 11.9 GB Fits | 12.6 GB Fits | 15.4 GB Fits | 892 |
| 6-bit (Q6_K) • | 13.8 GB Fits | 13.9 GB Fits | 14.6 GB Fits | 17.5 GB Fits | 752 |
| 8-bit (Q8_0) • | 16.3 GB Fits | 16.4 GB Fits | 17.1 GB Fits | 19.9 GB Fits | 633 |
| 16-bit (F16/BF16) | 30.0 GB Tight | 30.1 GB Tight | 30.8 GB No | 33.6 GB No | 337 |
• real GGUF file size from mradermacher/DeepSeek-V2-Lite-Chat-GGUF. Speed is a bandwidth ceiling (1792 GB/s), not a benchmark. Methodology.
More
DeepSeek V2 Lite Chat on every GPU and Mac · Every model that runs on the RTX 5090 · Other models that fit 32 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (deepseek-ai/DeepSeek-V2-Lite-Chat), NVIDIA specs. Data as of 2026-11-17.