Job PlannerDescribe the job. Get the order.
Say what you want to run. The planner sizes it: memory, compute, the GPU classes that fit, the hours, and the cost at live open-market prices. One click turns the result into a sealed order.
Run inferencegenerate or embedFine-tune 路 LoRAadapters, 16-bit baseFine-tune 路 QLoRAadapters, 4-bit baseTrain 路 fullevery weight updates
Memory needed23 GB
Compute1.6 脳 1018floating-point operations
Cheapest fit$1.691 脳 RTX 3090 for 9 h
| class | GPUs | work time | order | median price | cost | |
|---|---|---|---|---|---|---|
| RTX 3090 | 1 脳 24 GB | 8.9 h | 9 GPU路h | $0.187 | $1.69 | Seal this order |
| RTX 4090 | 1 脳 24 GB | 3.8 h | 4 GPU路h | $0.443 | $1.77 | Seal this order |
| RTX 5090 | 1 脳 32 GB | 3.0 h | 4 GPU路h | $0.561 | $2.24 | Seal this order |
| L40S | 1 脳 48 GB | 3.5 h | 4 GPU路h | $0.734 | $2.94 | Seal this order |
| A100 SXM4 | 1 脳 80 GB | 4.1 h | 5 GPU路h | $0.802 | $4.01 | Seal this order |
| H100 PCIe | 1 脳 80 GB | 1.7 h | 2 GPU路h | $2.470 | $4.94 | Seal this order |
| H100 SXM | 1 脳 80 GB | 1.3 h | 2 GPU路h | $3.181 | $6.36 | Seal this order |
| B200 | 1 脳 192 GB | 34 min | 1 GPU路h | $7.813 | $7.81 | Seal this order |
| H200 | 1 脳 141 GB | 1.3 h | 2 GPU路h | $5.001 | $10.00 | Seal this order |