MiniMax M2.7
MiniMaxAI/MiniMax-M2.7
Parameters
230B
Context
200K
Downloads
861.6K
Likes
1.2K
Architecture
moe_transformer
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 546.9 GB RAM 820.3 · Disk 707.3 | 587.3 GB RAM 880.9 · Disk 733.3 | 708.8 GB RAM 1063.2 · Disk 772.3 |
| FP88bit | 275.9 GB RAM 413.9 · Disk 373.2 | 304.0 GB RAM 456.0 · Disk 399.2 | 413.2 GB RAM 619.8 · Disk 438.2 |
| INT88bit | 275.9 GB RAM 413.9 · Disk 373.2 | 304.0 GB RAM 456.0 · Disk 399.2 | 413.2 GB RAM 619.8 · Disk 438.2 |
| AWQ4bit | 145.5 GB RAM 218.3 · Disk 212.3 | 167.7 GB RAM 251.5 · Disk 238.3 | 270.9 GB RAM 406.4 · Disk 277.3 |
| GPTQ4bit | 147.2 GB RAM 220.8 · Disk 214.4 | 169.4 GB RAM 254.1 · Disk 240.4 | 272.8 GB RAM 409.2 · Disk 279.4 |
| GGUF4bit | 150.6 GB RAM 225.9 · Disk 218.6 | 173.0 GB RAM 259.5 · Disk 244.6 | 276.5 GB RAM 414.7 · Disk 283.6 |
GPU Recommendations
Best ValueBEST
4× H200 SXM
564 GB VRAM
$10,220/mo
≈ 103.7 tok/s
Effective production throughput ≈ 72.6 tok/s
score 0.3486
Cheapest
8× RTX 3090
192 GB VRAM
$630/mo
≈ 27 tok/s
Effective production throughput ≈ 18.9 tok/s
score 0.5354
Performance
8× H200 SXM
1128 GB VRAM
$20,440/mo
≈ 138.2 tok/s
Effective production throughput ≈ 96.7 tok/s
score 0.2197
Min Complexity
1× B200
180 GB VRAM
$3,013/mo
≈ 0 tok/s (N/A — benchmark unavailable)
Effective production throughput ≈ 0 tok/s
score 0.5789
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
B200
180 GB VRAM
$3,013
/mo · vast
B200 SXM
180 GB VRAM
—