Radeon Instinct MI200
AMD
64GB VRAM · 1640 GB/s bandwidth · 181.0 FP16 TFLOPS · 300W TDP
The Radeon Instinct MI200 has 64GB of VRAM with 1640 GB/s memory bandwidth and 181.0 TFLOPS FP16 compute. At Q4 quantization, it can comfortably run Gemma 3 4B (533 tok/s), Qwen 2.5 7B (280 tok/s), Llama 3.1 8B (265 tok/s). Models larger than ~109B parameters won't fit even at Q4. Electricity cost is approximately $32/month at 300W TDP.
What LLMs can you run?
| Model | Params | Q4 Weight | Fit | Decode |
|---|---|---|---|---|
| Gemma 3 4B | 4.0B | 2 GB | comfortable | 533 tok/s |
| Qwen 2.5 7B | 7.6B | 4 GB | comfortable | 280 tok/s |
| Llama 3.1 8B | 8.0B | 4 GB | comfortable | 265 tok/s |
| Mistral Small 24B | 24.0B | 12 GB | comfortable | 88 tok/s |
| Gemma 3 27B | 27.4B | 14 GB | comfortable | 77 tok/s |
| Qwen 2.5 Coder 32B | 32.5B | 16 GB | comfortable | 65 tok/s |
| Llama 3.3 70B | 70.6B | 35 GB | comfortable | 30 tok/s |
| Qwen 2.5 72B | 72.7B | 36 GB | tight | 29 tok/s |
| Llama 3.1 405B | 405B | 202 GB | won't fit | |
| DeepSeek R1 671B | 671B | 336 GB | won't fit |
Similar GPUs
| GPU | VRAM | BW | TFLOPS | TDP |
|---|---|---|---|---|
| Radeon Instinct MI210 | 64GB | 1640 GB/s | 181.0 | 300W |
| H100 SXM5 64 GB | 64GB | 2020 GB/s | 267.6 | 700W |
| M1 Ultra 64GB | 64GB | 800 GB/s | 42.5 | 120W |
| M2 Ultra 64GB | 64GB | 800 GB/s | 54.4 | 120W |
| M4 Max 64GB | 64GB | 546 GB/s | 36.9 | 75W |
Compare with another GPU
Select another GPU to compare specs and model performance side by side.