Loading…
Loading…
Models / K2-Think
GPU memory needed (smallest version of each): K2-Think ~32.8B about 77.4 GB. The cheapest runs on 4x RTX A5000 from $1.01/h right now.
1 model (~32.8B) · ↓ 237 downloads a month · from $1.01/h on 4x RTX A5000 right now. Each model's versions do the same job; quantized ones (AWQ, FP8, 4-bit) need less GPU memory and run on cheaper machines.
| Version | Runtime | Weights | GPU memory | Cheapest machine now |
|---|---|---|---|---|
| OriginalIFM/K2-Think | vLLM | 66 GB | 77.4 GB | $1.01/h 4x RTX A5000 |
Model details come from Hugging Face and are refreshed daily; prices follow the live offers. GPU memory is GridPup's estimate: the weights, room for an 8,192-token conversation (chat models) and the runtime's own memory. Large models are split across the GPUs of one machine ("4x RTX 4090").