Skip to main content
Cached models are currently in beta. Join our Discord if you’d like to provide feedback.
Enabling cached models on your Serverless endpoint can reduce cold start times to just a few seconds and dramatically reduce the cost for loading large models.

​
Why use cached models?

  • Faster cold starts: A “cold start” refers to the delay between when a request is received by an endpoint with no running workers and when a worker is fully “warmed up” and ready to handle the request. Using cached models can reduce cold start times to just a few seconds, even for large models.
  • Reduced costs: You aren’t billed for worker time while your model is being downloaded. This is especially impactful for large models that can take several minutes to load.
  • Accelerated deployment: You can deploy cached models instantly without waiting for external downloads or transfers.
  • Smaller container images: By decoupling models from your container image, you can create smaller, more focused images that contain only your application logic.
  • Shared across workers: Multiple workers running on the same host machine can reference the same cached model, eliminating redundant downloads and saving disk space.

​
Cached model compatibility

Cached models work with any model hosted on Hugging Face, including:
  • Public models: Any publicly available model on Hugging Face.
  • Gated models: Models that require you to accept terms (provide a Hugging Face access token).
  • Private models: Private models your Hugging Face token has access to.
Cached models aren’t suitable if your model is private and not hosted on Hugging Face. In that case, bake it into your Docker image instead.

​
How it works

When you select a cached model for your endpoint, Runpod automatically tries to start your workers on hosts that already contain the selected model. If no cached host machines are available, the system delays starting your workers until the model is downloaded onto the machine where your workers will run, ensuring you still won’t be charged for the download time.

​
Access cached models in your application

Cached models are stored at /runpod-volume/huggingface-cache/hub/ following Hugging Face cache conventions. The directory structure replaces forward slashes (/) from the original model name with double dashes (--), and includes a version hash subdirectory.
While cached models use the same mount path as network volumes (/runpod-volume/), the model loaded from the cache will load significantly faster than the same model loaded from a network volume.
The path structure follows this pattern: 
/runpod-volume/huggingface-cache/hub/models--HF_ORGANIZATION--MODEL_NAME/snapshots/VERSION_HASH/
For example, the model gensyn/qwen2.5-0.5b-instruct would be stored at: 
/runpod-volume/huggingface-cache/hub/models--gensyn--qwen2.5-0.5b-instruct/snapshots/317b7eb96312eda0c431d1dab1af958a308cb35e/
If your application requires specific paths, configure it to scan /runpod-volume/huggingface-cache/hub/ for models.

​
Enable cached models

Follow these steps to select and add a cached model to your Serverless endpoint:
1

Create a new endpoint

Navigate to the Serverless section of the console and click New Endpoint.
2

Configure the model

In the Endpoint Configuration step, scroll down to Model (optional) and add the link for the model you want to use.For example, https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct.
3

Add an access token (if needed)

If you’re using a gated model, you’ll need to enter a Hugging Face access token.
4

Deploy the endpoint

Complete your endpoint configuration and click Deploy Endpoint .
You can add a cached model to an existing endpoint by selecting Manage → Edit Endpoint in the endpoint details page and updating the Model (optional) field.

​
Current limitations

If a Hugging Face repository contains multiple quantization versions of a model (for example, 4-bit AWQ and 8-bit GPTQ versions), the system currently downloads all quantization versions. The ability to select specific quantizations will be available in a future update.