Can I run this LLM?
Can a Mac Studio M5 Ultra (512GB) run Qwen3.8 Flash Next (MoE)?
Qwen · See model card · estimated for a 512.0 GB machine at 4K context.
Yes — it runs great.
Runs greatBest fit: Qwen3.8 Flash Next (MoE) at Q8_0 · 188.3 GB weights · 73–110 tok/s on a Mac Studio M5 Ultra (512GB).
Arena Text Overall · style-controlled
UnratedAmbiguous matchAmbiguous Arena identity; unrated
Arena identity is ambiguous. No exact configuration is present in the latest Overall entry set. No related-family rating is inherited. No rating is assigned or used for ranking. Local quantization quality is unmeasured.
Snapshot published 2026-10-08 · snapshot imported 2026-10-11 UTC · validated last-good data · manual refresh
Arena leaderboard dataset · Arena / lmarena-ai, Arena Leaderboard Dataset (CC BY 4.0). Selected Text Overall style-controlled rows; field names adapted, ratings unchanged. · dataset CC BY 4.0 · revision e16f90232986
- Total memory at 4K context
- 189.2 GB
- Weights + KV cache + runtime overhead
- Estimated generation
- 73–110 tok/s
- Estimated first token
- ~1.9s
Every quantization on this machine
On small screens, each quant lists every metric below its name. On wider screens, scroll sideways for total memory, speed and fit. Keyboard: focus the table, then use the arrow keys. Quant names stay visible.
| Quant | Weights | Total memory | Speed | Fit |
|---|---|---|---|---|
| Q2_K | 80.9 GB | 81.8 GB | 108–162 tok/s | Runs great |
| Q3_K_M | 92.0 GB | 92.9 GB | 103–154 tok/s | Runs great |
| Q4_K_M | 119.6 GB | 120.5 GB | 92–138 tok/s | Runs great |
| Q5_K_M | 134.7 GB | 135.6 GB | 87–131 tok/s | Runs great |
| Q6_K | 168.1 GB | 169.0 GB | 78–117 tok/s | Runs great |
| Q8_0 | 188.3 GB | 189.2 GB | 73–110 tok/s | Runs great |
| FP16 | 354.0 GB | 354.9 GB | 49–73 tok/s | Tight / slow |
Ran Qwen3.8 Flash Next (MoE) on a Mac Studio M5 Ultra (512GB)? Add your real numbers →