Supported Models
Learn about StarOps Supported Models
StarOps allows you to host and deploy several open source large language models for text generation from your own VPC. You can see our list of supported models below. If there is a model you’d like us to support please let us know by sending us feedback.
As you will be hosting these models from your own VPC, it is your responsibility to ensure that your service and/or application adheres to the acceptable use policy of each model provider, as well as any regulations governing your product and/or service in the regions you operate in.
| Model Name | Size | Instance Type | vLLM CPU Request | vLLM CPU Limit | vLLM Memory Request | vLLM Memory Limit | vLLM GPU Request | vLLM GPU Limit |
|---|---|---|---|---|---|---|---|---|
| Gemma-3-1B-it | 1B | g6.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| Llama-3.2-3B-Instruct | 3B | g5.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| Llama-3.1-8B-Instruct | 8B | g5.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| Llama-3.1-Nemotron-Nano-8B-v1 | 8B | g5.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| OPT | 125M | g6.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| Phi-4-mini-instruct | 3.8B | g6.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |
| Qwen-2.5-7B-Instruct | 7B | g5.xlarge | 2 vCPU | 4 vCPU | 8 GiB | 16 GiB | 1 | 1 |