C92.5 tok/s decode
~$10,000 MSRP
Can it run?
Runs well
Using Q5_K_M in llama.cpp
Fit status
Runs well
Decode
45.9 tok/s
TTFT
4218 ms
Safe context
8K
Memory
21.6 GB / 32.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 42.7 tok/s | 6593 ms | 8K |
| Chat | C | Runs well | 42.7 tok/s | 2472 ms | 8K |
| Coding | C | Runs well | 45.9 tok/s | 4218 ms | 8K |
| RAG | C | Runs well | 42.7 tok/s | 8241 ms | 8K |
| Reasoning | C | Runs well | 42.7 tok/s | 5357 ms | 8K |
How InternLM 20B (20B params) fits at each quantization level on NVIDIA V100 32GB (32.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 7.8 GB | Low | D34 |
Q3_K_S | 3 | 9.8 GB | Low | D36 |
NVFP4 | 4 |
huggingface-cli download internlm-20bUpgrade options
~$10,000 MSRP
11.2 GB |
| Medium |
| D37 |
Q4_K_M | 4 | 12.2 GB | Medium | D37 |
Q5_K_M | 5 | 14.4 GB | High | D39 |
Q6_K | 6 | 16.4 GB | High | C40 |
Q8_0Best for your GPU | 8 | 21.4 GB | Very High | C44 |
F16 | 16 | 41.0 GB | Maximum | F0 |