deepseek-ai/DeepSeek-R1-Distill-Llama-8B

vRam: 24 GB

Disk: 18GB

Context window: 20k - 100k tokens

AWS: G6.Xlarge

Runpod: $0.22

On Prem: RTX 4090

--host 0.0.0.0 --port 8000 --model deepseek-ai/DeepSeek-R1-Distill-Llama-8B --tensor-parallel-size 1 --max-model-len 20000 --enforce-eager

ai21labs/AI21-Jamba-1.5-Mini

vRAM: 80 GB

Context Window: 100k Tokens

Disk: 110 GB

Runpod cost: $0.80

AWS: Unfeasible

On Prem: 1x A100 80GB / 4 x RTX 4000 ADA

--host 0.0.0.0 --port 8000 --model ai21labs/AI21-Jamba-1.5-Mini --tensor-parallel-size 4 --max-model-len 100000 --quantization experts_int8