# GPUtopia/Arguflow bounty integration

## Requirements

- [X] OpenAI compliant API over an LLM thru [LocalAI](https://github.com/arguflow/LocalAI) running on gputopia network
- [X] Text-embedding model running in [Arguflow embedding server](https://github.com/arguflow/arguflow/tree/main/embedding-server) on gputopia network
- [ ] ~10-minute deployment of Arguflow on top of GPUTopia compute

## Bounty amount

## Proposed path forward for gputopia

- advanced routing and model preloading
- always picking handler that have the model loaded
- sending out preloads in anticipation of future requests

- check fastembed/onyx vs torch for exe size and performance esp parallelism, one of these should work
- bge-large-en-v1.5
- ember-v1
- gte-large
- stella-base-en-v2

## For Arguflow
- please test our api endpoint (sign up get an api token, and try it out)
- send feeback and requests until it works nicely