GPU Acceleration
As of Typesense Server v0.25.0, Typesense can generate embeddings natively, with data in your JSON documents, using any of the built-in ML models listed here (or using OpenAI API or PaLM API).
When you use one of the built-in ML models, you can improve performance of the embedding generation significantly, during indexing and at search time (for eg, when doing semantic / hybrid search) by having Typesense utilize a GPU.
GPU Acceleration is available in the following RAM / vCPU configuration in select regions:
| Memory | vCPU |
|---|---|
| 8 GB | 4 vCPUs |
| 16 GB | 4 vCPUs |
| 16 GB | 8 vCPUs |
| 32 GB | 8 vCPUs |
| 32 GB | 16 vCPUs |
| 64 GB | 16 vCPUs |
| 64 GB | 32 vCPUs |
| 128 GB | 32 vCPUs |
| 128 GB | 64 vCPUs |
| 192 GB | 48 vCPUs |
| 256 GB | 64 vCPUs |
| 384 GB | 96 vCPUs |