Can I run this LLM?
Can a AMD Instinct MI300X (192GB) run AesCode 32B (text-only)?
AesCode · Apache 2.0 · estimated for a 192.0 GB machine at 4K context.
Yes — it runs great.
Runs greatBest fit: AesCode 32B (text-only) at FP16 · 65.5 GB weights · 51–76 tok/s on a AMD Instinct MI300X (192GB).
Pack & allocation notes: Text-only estimates exclude the vision projector and image-input runtime memory. The FP16 label uses supplied BF16 weights.
Arena Text Overall · style-controlled
UnratedNo exact matchNo reviewed exact Arena match
No reviewed exact model/configuration match in the Arena Text Overall style-controlled snapshot. No family or editorial score is substituted. 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
- 67.4 GB
- Weights + KV cache + runtime overhead
- Estimated generation
- 51–76 tok/s
- Estimated first token
- ~1.6s
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 | 13.0 GB | 14.9 GB | 158–237 tok/s | Runs great |
| Q3_K_M | 16.5 GB | 18.4 GB | 138–208 tok/s | Runs great |
| Q4_K_M | 20.5 GB | 22.4 GB | 121–182 tok/s | Runs great |
| Q5_K_M | 23.8 GB | 25.7 GB | 110–165 tok/s | Runs great |
| Q6_K | 28.0 GB | 29.9 GB | 99–148 tok/s | Runs great |
| Q8_0 | 34.8 GB | 36.7 GB | 84–126 tok/s | Runs great |
| FP16 | 65.5 GB | 67.4 GB | 51–76 tok/s | Runs great |
Ran AesCode 32B (text-only) on a AMD Instinct MI300X (192GB)? Add your real numbers →