What is this?

When you try to embed a file in AnythingLLM, you might see a "Fetch failed" error. There are a few reasons why this might happen and all of them are fixable quite easily and are all related to the machine running AnythingLLM or firewall permissions.

Below are the most common fixes for this error ordered from the most likely to the least likely.

Check if the machine running AnythingLLM is blocking downloads from HuggingFace or AWS.

This error applies to you if:

• You are using the default AnythingLLM embedder model
• You may have a firewall blocking downloads from HuggingFace or AWS either by default or because you have a custom firewall installed by whoever manages your network.

Why is this happening?

This error happens when the machine running AnythingLLM is blocking downloads from HuggingFace or AWS. We do not pre-bundle the embedding model into the app, so the machine needs to download the model for its very first use. After it is downloaded, the model is cached so it doesn't need to be downloaded again. Your embeddings for the default embedder model are always done locally, this is just a problem with downloading the model GGUF and tokenizer.

How to fix it?

1. Check your storage folder and see if a folder named models/Xenova exists.

• If this folder does not exist, it's likely that the machine is blocking downloads from HuggingFace or AWS.
• Unblock the huggingface.co and api.huggingface.co domains on your machine.
• Try embedding again.
• Unblock this origin: https://s3.us-west-1.amazonaws.com/public.useanything.com/support/models/
• Try embedding again.

Still not working? Try the next solution.

Windows Visual C++ Redistributable

This error applies to you if:

• You are using the default AnythingLLM embedder model
• You are on Windows

Why is this happening?

This error happens when the machine running AnythingLLM is missing the Windows Visual C++ Redistributable. This is a library that is required to run the model.

How to fix it?

1. Download the Visual C++ Redistributable v14.x (opens in a new tab) and install it.
2. Try embedding again.

Still not working? Try the next solution.

Your CPU is not supported

This error applies to you if:

• You are using the default AnythingLLM vector database

Why is this happening?

LanceDB (opens in a new tab) is a vector database that is used to store the embeddings. It is the default vector database for AnythingLLM.

Your CPU is not supported if you are using a CPU that does not support AVX2.

How to fix it?

1. Use a machine with a supported CPU.
2. Use another vector database provider for vector storage. We support most of the popular vector databases.