Mac Studio M1 Ultra 64GBBudget pick
C103 tok/s decode
~$3,999 MSRP
Can it run?
Runs well
Using Q4_K_M in Ollama
Fit status
Runs well
Decode
262.2 tok/s
TTFT
738 ms
Safe context
61K
Memory
10.6 GB / 40.0 GB
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Agentic Coding | C | Runs well | 305.9 tok/s | 921 ms | 110K |
| Chat | C | Runs well | 305.9 tok/s | 350 ms | 31K |
| Coding | C | Runs well | 262.2 tok/s | 738 ms | 61K |
| RAG | C | Runs well | 305.9 tok/s | 1151 ms | 110K |
| Reasoning | C | Runs well | 305.9 tok/s | 748 ms | 61K |
How Mistral 7B Instruct v0.2 (7B params) fits at each quantization level on NVIDIA A100 40GB (40.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 2.7 GB | Low | D31 |
Q3_K_S | 3 | 3.4 GB | Low | D31 |
NVFP4 | 4 |
Upgrade options
~$3,999 MSRP
~$3,999 MSRP
| Medium |
| D32 |
Q4_K_M | 4 | 4.3 GB | Medium | D32 |
Q5_K_M | 5 | 5.0 GB | High | D32 |
Q6_K | 6 | 5.7 GB | High | D32 |
Q8_0 | 8 | 7.5 GB | Very High | D33 |
F16Best for your GPU | 16 | 14.3 GB | Maximum | D37 |
~$3,999 MSRP