Documentation
Coming soon
Custom-model serving
Bring your own model, checkpoint or fine-tuned model to Arnict.
Run a model of your choice
Coming soon
Custom-model inference is planned through Arnict’s dedicated GPU offering.
Bring a custom model you have chosen or fine-tuned and use it for your application. This includes checkpoints produced with LoRA or full fine-tuning; you do not need to limit your choice to the public Arnict catalog.
The deployment guide will explain model submission, supported formats, configuration and API access before this feature opens.
- Dedicated GPU inferenceReview the planned deployment and pricing model.
- LoRA and full fine-tuningPrepare for adapting a base model to your task.
Plan your deployment
- Confirm that you have the rights to run and use the model.
- Identify the model format, required capabilities and expected traffic.
- Keep a small evaluation set so you can verify output after deployment.
- Review the published configuration and pricing before starting a deployment.
Custom-model serving is not currently available and no custom-serving price is active. Join the waitlist for access updates.
- Get deployment updatesJoin the dedicated-inference waitlist.