M5 Ultra 96GB vs AMD Strix Halo 128GB for local LLMs
12 reports on the M5 Ultra 96GB and 348 on the AMD Strix Halo 128GB
Which is faster for local LLMs?
On Qwen3.8 27B at 8-bit, the M5 Ultra 96GB has 15.0 generation tokens/s (one run); the AMD Strix Halo 128GB has 7.5 generation tokens/s (3 runs); prompt processing 257.8 tokens/s (2 runs). Those are community reports whose engines and settings differ, so the gap is a rough sign, not a controlled measure of the two cards. Memory bandwidth, which caps how fast a card can write, is 4.8x higher on the M5 Ultra 96GB (1,229 GB/s on the M5 Ultra 96GB, 256 on the AMD Strix Halo 128GB).
- Memory
- 96GB unified
- Memory bandwidth
- 1,229 GB/s
- FP16 compute
- not on record
- Reports
- 12
- Memory
- 128GB unified
- Memory bandwidth
- 256 GB/s
- FP16 compute
- 59 TFLOPS
- Reports
- 348
Measured on both
Generation and prompt processing speeds from plain runs (one device, one request, no speculative decoding, the whole model in the device's memory) of the same model size at the same bit class. Each side shows the engines and reported context behind its median. Context can mean a configured window or prompt depth, so these are community comparisons with differing settings. How to read these
| Model | M5 Ultra 96GB | AMD Strix Halo 128GB |
|---|---|---|
| Qwen3.8 27B8-bit | 15.0one runengine not statedcontext not stated | 7.53 runsPP 257.8 (2 runs)llama.cpp, 1 not statedcontext not stated |
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape. Where a model was also measured above, the measurement wins.
| Model | M5 Ultra 96GB | AMD Strix Halo 128GB |
|---|---|---|
| Qwen3.8 27B | 14.9 t/sFP16 | 3.5 t/sFP16 |
| Qwen3.8 125B · 6B active | 103.1 t/sQ3_K_M | 12.6 t/sQ6_K |
| Qwen3.6 35B · 3B active | 107.0 t/sQ8_0 | 12.6 t/sFP16 |
| DeepSeek V4 Flash 284B · 13B active | doesn't fit at these settings | 33.5 t/sQ2_K |
| DeepSeek V4.1 Flash 552B · 16B active | doesn't fit at these settings | doesn't fit at these settings |
| Qwen3.6 27B | 14.9 t/sFP16 | 3.5 t/sFP16 |
| Gemma 4 26B · 4B active | 35.5 t/sFP16 | 9.5 t/sFP16 |
Change the context, RAM or model in the calculator: M5 Ultra 96GB · AMD Strix Halo 128GB
Every report on each card, with its source: M5 Ultra 96GB · AMD Strix Halo 128GB