Coding AIMIT
Devstral 2 123B
mistralai/Devstral-2-123B
Parameters
123B
Context
128K
Downloads
0
Likes
0
Architecture
transformer
GPU Requirements
VRAM / RAM / disk estimates by quantization and usage scenario
| Quantization | Minimum | Recommended | Production |
|---|---|---|---|
| FP1616bit | 294.8 GB RAM 442.2 · Disk 396.4 | 323.7 GB RAM 485.5 · Disk 422.4 | 433.8 GB RAM 650.7 · Disk 461.4 |
| FP88bit | 149.9 GB RAM 224.8 · Disk 217.7 | 172.2 GB RAM 258.3 · Disk 243.7 | 275.7 GB RAM 413.5 · Disk 282.7 |
| INT88bit | 149.9 GB RAM 224.8 · Disk 217.7 | 172.2 GB RAM 258.3 · Disk 243.7 | 275.7 GB RAM 413.5 · Disk 282.7 |
| AWQ4bit | 80.1 GB RAM 120.2 · Disk 131.7 | 99.3 GB RAM 148.9 · Disk 157.7 | 199.6 GB RAM 299.4 · Disk 196.7 |
| GPTQ4bit | 81.0 GB RAM 121.5 · Disk 132.8 | 100.2 GB RAM 150.4 · Disk 158.8 | 200.6 GB RAM 300.9 · Disk 197.8 |
| GGUF4bit | 82.8 GB RAM 124.3 · Disk 135.1 | 102.1 GB RAM 153.2 · Disk 161.1 | 202.6 GB RAM 303.9 · Disk 200.1 |
GPU Recommendations
Best ValueBEST
4× H200 SXM
564 GB VRAM
$10,220/mo
≈ 193.9 tok/s
Effective production throughput ≈ 135.7 tok/s
score 0.3163
Cheapest
8× Tesla V100
128 GB VRAM
$79.4/mo
≈ 0 tok/s (N/A — benchmark unavailable)
Effective production throughput ≈ 0 tok/s
score 0.5515
Performance
8× H200 SXM
1128 GB VRAM
$20,440/mo
≈ 258.5 tok/s
Effective production throughput ≈ 181 tok/s
score 0.2197
Min Complexity
1× H200 SXM
141 GB VRAM
$2,555/mo
≈ 80.8 tok/s
Effective production throughput ≈ 56.6 tok/s
score 0.6588
💬 Community — real deployment experience
0 Articles · 0 Benchmarks
GPUs that fit (single card)
H200 SXM
141 GB VRAM
$2,555
/mo · lambda
B200
180 GB VRAM
$3,013
/mo · vast
B200 SXM
180 GB VRAM
—