Ryzen AI Max+ 395 (Radeon 8060S) vs Ryzen AI Max 390 (Radeon 8050S) for local AI
The Ryzen AI Max+ 395 (Radeon 8060S) fits more: up to gpt-oss 120B, against Llama 3.3 70B on the Ryzen AI Max 390 (Radeon 8050S). They generate at about the same speed, since their memory bandwidth is close (256 and 256 GB/s).
Memory and bandwidth
Memory decides which models fit. Bandwidth decides how fast they generate, because every token reads the model's active weights once.
| Ryzen AI Max+ 395 (Radeon 8060S) | Ryzen AI Max 390 (Radeon 8050S) | |
|---|---|---|
| Memory | 64 or 96 or 128 GB unified memory | 32 or 64 GB unified memory |
| Usable for models | About 96 GB of 128 GB | About 48 GB of 64 GB |
| Bandwidth | 256 GB/s | 256 GB/s |
| Biggest model that fits | gpt-oss 120B | Llama 3.3 70B |
| Best everyday pick | gpt-oss 120B, ~31 tok/s | Qwen3 30B-A3B, ~45 tok/s |
Every model, side by side
Estimated generation speed at the largest memory size each was sold in (128 GB and 64 GB).
| Model | Ryzen AI Max+ 395 (Radeon 8060S) | Ryzen AI Max 390 (Radeon 8050S) |
|---|---|---|
| Qwen3 4B About 4 GB | ~56 tok/s Runs great | ~56 tok/s Runs great |
| Qwen3 8B About 7 GB | ~27 tok/s Usable | ~27 tok/s Usable |
| Gemma 3 12B About 10 GB | ~17 tok/s Usable | ~17 tok/s Usable |
| Qwen3 14B About 11 GB | ~15 tok/s Usable | ~15 tok/s Usable |
| gpt-oss 20B About 16 GB | ~45 tok/s Runs great | ~45 tok/s Runs great |
| Gemma 3 27B About 19 GB | ~8 tok/s Slow | ~8 tok/s Slow |
| Qwen3 30B-A3B About 21 GB | ~45 tok/s Runs great | ~45 tok/s Runs great |
| Qwen3 32B About 22 GB | ~7 tok/s Slow | ~7 tok/s Slow |
| Llama 3.3 70B About 45 GB | ~3 tok/s Slow | ~3 tok/s Slow |
| gpt-oss 120B About 67 GB | ~31 tok/s Runs great | Needs 67 GB |
| Qwen3 235B-A22B About 144 GB | Needs 144 GB | Needs 144 GB |
| DeepSeek R1 671B About 406 GB | Needs 406 GB | Needs 406 GB |
The Ryzen AI Max+ 395 (Radeon 8060S): it fits bigger models and runs faster. Price matters as much as either number: check used prices on the builds page, where every parts list has a measured or estimated speed.
Both columns use the same formula: memory bandwidth divided by the weights read per token, calibrated on our own rack, with Ollama's default 4-bit sizes. Real results are usually within about a third. Measured either one? Send us the number.
Other matchups
- Ryzen AI Max+ 395 (Radeon 8060S) vs NVIDIA DGX Spark128 GB · 273 GB/s
- Ryzen AI Max+ 395 (Radeon 8060S) vs Apple M4 Max36 / 48 / 64 / 128 GB · 546 GB/s
- Ryzen AI Max 390 (Radeon 8050S) vs Apple M1 Max32 / 64 GB · 400 GB/s
- Ryzen AI Max 390 (Radeon 8050S) vs Apple M4 Pro24 / 48 / 64 GB · 273 GB/s
- Ryzen AI Max 390 (Radeon 8050S) vs Apple M5 Pro24 / 48 / 64 GB · 307 GB/s
- Ryzen AI Max 390 (Radeon 8050S) vs Jetson AGX Orin64 / 32 GB · 205 GB/s