Choosing the Right Model
| Model | Best for | VRAM | Est. cost/mo |
|---|---|---|---|
| Llama 3.2 3B | Assistants, customer support, chatbots | 8 GB | $28 |
| DeepSeek R1 7B | Agents, research, complex reasoning | 16 GB | $55 |
| Phi-3 Mini 4K | Lightweight tasks, low latency | 8 GB | $28 |
Costs are AWS infrastructure estimates billed directly by your cloud provider — not by CaseDesk.