▸ DEVICE UNDER TEST
AMD Instinct MI300X — 192 GB VRAM.
▸ INSTINCT MI300X SPEC
- BRAND
- AMD
- VRAM
- 192 GB HBM3
- BANDWIDTH
- 5300 GB/s
- FP16 COMPUTE
- 1307.4 TFLOPS
- FP32 COMPUTE
- 163.4 TFLOPS
- TDP
- 750 W
- ARCHITECTURE
- CDNA3
- MSRP
- $15000
▸ AI CAPABILITY
307/ 331 models @ Q4
With 192 GB VRAM and 5300 GB/s bandwidth, this GPU handles models up to 236B parameters.
Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~606 tok/s.
§ 01TOP MODELS FOR INSTINCT MI300X
307 FIT · SHOWING 20| MODEL | SIZE | VRAM Q4 | TOK/S | AVG |
|---|---|---|---|---|
| DeepSeek V2 236B | 236B | 144.7 GB | 224 | 40.4 |
| DeepSeek-V2.5 236B | 236B | 144.7 GB | 224 | 64.0 |
| DeepSeek-Coder-V2 236B | 236B | 144.7 GB | 224 | 61.8 |
| Qwen3-235B-A22B | 235.1B | 144.2 GB | 214 | 55.9 |
| MiniMax-M2.5 228.7B | 228.7B | 140.3 GB | 224 | 62.6 |
| Falcon 180B | 180B | 110.5 GB | 26 | 51.3 |
| bloom 176.2B | 176.2B | 108.2 GB | 27 | 15.0 |
| dots.llm1.inst 142.8B | 142.8B | 87.8 GB | 33 | — |
| WizardLM 2 8x22B | 141B | 86.7 GB | 121 | 42.4 |
| Mixtral-8x22B | 140.6B | 86.4 GB | 120 | 31.9 |
| DBRX 132B | 132B | 81.2 GB | 131 | 46.3 |
| Qwen3.5-122B-A10B | 125.1B | 77.0 GB | 38 | 45.5 |
| Pixtral Large 124B | 124B | 76.3 GB | 38 | 39.3 |
| Mistral-Large 123B | 123B | 75.7 GB | 38 | 33.5 |
| Devstral 2 123B | 123B | 75.7 GB | 38 | 38.1 |
| Qwen 3.5 122B A10B | 122B | 75.1 GB | 471 | 56.8 |
| Nemotron 3 Super 120B | 120B | 73.8 GB | 393 | 57.3 |
| Nemotron 3 Super 120B-A12B | 120B | 73.8 GB | 393 | 53.2 |
| Mistral Small 4 119B | 119B | 73.2 GB | 725 | 50.2 |
| GPT-OSS 120B | 117B | 72.0 GB | 924 | 54.1 |