Minisforum MS-S1 MAX (Ryzen AI Max+ 395) vs Corsair AI Workstation 300 (Ryzen AI Max+ 395)
The cheapest Minisforum MS-S1 MAX (Ryzen AI Max+ 395) lists for about $99 more than the cheapest Corsair AI Workstation 300 (Ryzen AI Max+ 395). Rows where the two differ are tinted.
AMD unified memory
Minisforum MS-S1 MAX (Ryzen AI Max+ 395)
A Strix Halo mini workstation with 128GB, a PCIe slot, dual 10GbE and 80Gbps USB4 v2, designed to stand on a desk or mount in a rack. The PCIe slot leaves room for a graphics card later.
gpt-oss-20b, 4-bit: 73 tok/s
AMD unified memory
Corsair AI Workstation 300 (Ryzen AI Max+ 395)
Corsair's small-form-factor Strix Halo system with 128GB and Corsair's AI software bundle. Corsair markets it for local LLMs with up to 96GB assigned to the GPU.
gpt-oss-20b, 4-bit: 73 tok/s
| Specification | Minisforum MS-S1 MAX (Ryzen AI Max+ 395) | Corsair AI Workstation 300 (Ryzen AI Max+ 395) |
|---|---|---|
| Kind | AMD unified memory · Minisforum | AMD unified memory · Corsair |
| Chip or GPUs | Ryzen AI Max+ 395 | Ryzen AI Max+ 395 |
| Memory | 128 GB | 128 GB |
| Memory for models | 96 GB (GPU-usable memory) | 96 GB (GPU-usable memory) |
| Memory bandwidth | 256 GB/s | 256 GB/s |
| Price | ~$4,899 | ~$4,800 |
| Largest at 4-bit, 8k context | Nemotron 3 Super 120B-A12B | Nemotron 3 Super 120B-A12B |
| Software | ROCm, Vulkan, DirectML | ROCm, Vulkan, DirectML |
| Power | 320 W | 300 W |
What each one runs
Every model in the catalog that at least one of the two holds at 4-bit with 8,192 tokens of context. Copper speeds are measured medians; outlined ones are estimates. Rows where only one fits, or where the speed differs by more than a quarter, are tinted.
| Model | Minisforum MS-S1 MAX (Ryzen AI Max+ 395) | Corsair AI Workstation 300 (Ryzen AI Max+ 395) |
|---|---|---|
| Nemotron 3 Super 120B-A12B120B, 12B active · 81.0 GB file | 15 tok/smedian of 2 runs | 15 tok/smedian of 2 runs |
| Qwen3.5 122B-A10B122B, 10B active · 71.3 GB file | 20 tok/smedian of 2 runs | 20 tok/smedian of 2 runs |
| Mistral Medium 3.5 128B128B · 69.8 GB file | ~2.3–2.8 tok/sestimate | ~2.3–2.8 tok/sestimate |
| GLM-4.5-Air106B, 12B active · 68.0 GB file | ~15–19 tok/sestimate | ~15–19 tok/sestimate |
| Mistral Small 4 119B (2603)119B, 6.5B active · 67.6 GB file | ~30–39 tok/sestimate | ~30–39 tok/sestimate |
| Llama 4 Scout 17B-16E Instruct109B, 17B active · 60.9 GB file | 18 tok/smedian of 2 runs | 18 tok/smedian of 2 runs |
| gpt-oss-120b117B, 5.1B active · 58.5 GB file | 52 tok/smedian of 5 runs | 52 tok/smedian of 5 runs |
| Qwen3 Coder Next80B, 3B active · 45.2 GB file | ~66–85 tok/sestimate | ~66–85 tok/sestimate |
| Llama 3.3 70B Instruct70B · 39.6 GB file | 4.8 tok/smedian of 4 runs | 4.8 tok/smedian of 4 runs |
| Nemotron 3.5 Lightning 30B-A3B30B, 3B active · 23.7 GB file | ~47–60 tok/sestimate | ~47–60 tok/sestimate |
| Qwen3.6 35B-A3B35B, 3B active · 20.8 GB file | 60 tok/smedian of 7 runs | 60 tok/smedian of 7 runs |
| Ornith 1.5 35B-A3B35B, 3B active · 20.2 GB file | ~65–83 tok/sestimate | ~65–83 tok/sestimate |
| Qwen3 Coder 30B-A3B Instruct30.5B, 3.3B active · 17.3 GB file | ~60–77 tok/sestimate | ~60–77 tok/sestimate |
| Gemma 4 31B IT30.7B · 17.1 GB file | 11 tok/smedian of 2 runs | 11 tok/smedian of 2 runs |
| Granite 4.2 30B29.3B · 16.8 GB file | ~9.6–12 tok/sestimate | ~9.6–12 tok/sestimate |
| Qwen3.8 27B27B · 16.2 GB file | 12 tok/smedian of 2 runs | 12 tok/smedian of 2 runs |
| Gemma 4 26B-A4B IT25.2B, 3.8B active · 15.9 GB file | 52 tok/smedian of 2 runs | 52 tok/smedian of 2 runs |
| Qwen3.6 27B27B · 15.7 GB file | ~10–12 tok/sestimate | ~10–12 tok/sestimate |
| Devstral Small 2 24B (2512)24B · 13.3 GB file | ~12–15 tok/sestimate | ~12–15 tok/sestimate |
| gpt-oss-20b21B, 3.6B active · 10.8 GB file | 73 tok/smedian of 3 runs | 73 tok/smedian of 3 runs |
| Phi-414.7B · 8.3 GB file | ~20–23 tok/sestimate | ~20–23 tok/sestimate |
| Ministral 3 14B Reasoning (2512)13.9B · 7.7 GB file | ~21–25 tok/sestimate | ~21–25 tok/sestimate |
| Gemma 4 12B IT11.95B · 6.6 GB file | ~24–29 tok/sestimate | ~24–29 tok/sestimate |
| Qwen3.5 9B9B · 5.3 GB file | ~31–37 tok/sestimate | ~31–37 tok/sestimate |
| Granite 4.2 8B8.8B · 5.2 GB file | ~31–38 tok/sestimate | ~31–38 tok/sestimate |
| Gemma 4 E4B IT8B · 5.0 GB file | ~57–68 tok/sestimate | ~57–68 tok/sestimate |
| Qwen3 8B8.2B · 4.7 GB file | ~35–41 tok/sestimate | ~35–41 tok/sestimate |
| Llama 3.1 8B Instruct8B · 4.6 GB file | 40 tok/smedian of 2 runs | 40 tok/smedian of 2 runs |
| Qwen3.5 4B4B · 2.6 GB file | ~63–76 tok/sestimate | ~63–76 tok/sestimate |
Full pages: Minisforum MS-S1 MAX (Ryzen AI Max+ 395) · Corsair AI Workstation 300 (Ryzen AI Max+ 395) · check a different quant or context length