NVIDIA H200 141GBBudget pickB62.7 tok/s decode
~$30,000 MSRP
Can it run?
Runs well
Using Q3_K_M in Ollama
Fit status
Runs well
Decode
35.8 tok/s
TTFT
5401 ms
Safe context
22K
Memory
92.8 GB / 128.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Tight fit | 33.9 tok/s | 8319 ms | 37K |
| Chat | C | Runs well | 33.9 tok/s | 3119 ms | 12K |
| Coding | C | Runs well | 35.8 tok/s | 5401 ms | 22K |
| RAG | C | Tight fit | 33.9 tok/s | 10398 ms | 37K |
| Reasoning | C | Runs well | 33.9 tok/s | 6759 ms | 22K |
How Qwen3.5 122B A10B (122B params) fits at each quantization level on AMD Instinct MI250 128GB (128.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 47.6 GB | Low | D37 |
Q3_K_S | 3 | 59.8 GB | Low | D40 |
NVFP4 | 4 |
huggingface-cli download hf-unsloth--qwen3-5-122b-a10b-ggufUpgrade options
NVIDIA H200 141GBBudget pick~$30,000 MSRP
NVIDIA H200 PCIe 141GBBest value~$30,000 MSRP
68.3 GB |
| Medium |
| C41 |
Q4_K_M | 4 | 74.4 GB | Medium | C42 |
Q5_K_M | 5 | 87.8 GB | High | C44 |
Q6_KBest for your GPU | 6 | 100.0 GB | High | C45 |
Q8_0 | 8 | 130.5 GB | Very High | F0 |
F16 | 16 | 250.1 GB | Maximum | F0 |