Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
68.4 tok/s
TTFT
2829 ms
Safe context
24K
Memory
63.7 GB / 96.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | B | Runs well | 77.0 tok/s | 3658 ms | 41K |
| Chat | C | Runs well | 77.0 tok/s | 1372 ms | 13K |
| Coding | C | Runs well | 68.4 tok/s | 2829 ms | 24K |
| RAG | B | Runs well | 77.0 tok/s | 4573 ms | 41K |
| Reasoning | B | Runs well | 77.0 tok/s | 2972 ms | 24K |
How Solar Open 69B REAP i1 (69B params) fits at each quantization level on NVIDIA H20 96GB (96.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 26.9 GB | Low | D35 |
Q3_K_S | 3 | 33.8 GB | Low | D36 |
NVFP4 | 4 |
38.6 GB |
| Medium |
| D38 |
Q4_K_M | 4 | 42.1 GB | Medium | D38 |
Q5_K_M | 5 | 49.7 GB | High | C40 |
Q6_K | 6 | 56.6 GB | High | C42 |
Q8_0Best for your GPU | 8 | 73.8 GB | Very High | C44 |
F16 | 16 | 141.5 GB | Maximum | F0 |
