Fix Ollama model not found or pull failed

ollama run NAME normally downloads the model for you. A “model not found” or “pull model manifest: file does not exist” error almost always means the name or tag does not exist in the registry — a typo, the wrong quant tag, or a model type Ollama cannot load. Fix the name, then pull.

1. See what is installed

ollama list

Lists every model and tag on this machine; the ollama ls alias does the same.

2. Find the exact name and tag

Browse the model library at ollama.com/search; names are case-sensitive and a bare name means :latest.

ollama pull llama3.2

Pull the exact model:tag string from the library before you run it.

3. Pull a GGUF from Hugging Face

ollama run hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF

Runs any public GGUF repo directly; huggingface.co works as the host too.

ollama run hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q8_0

Append :QUANT to pick a quantization (the default is Q4_K_M), and the tag is case-insensitive.

4. Retry a failed pull

ollama pull hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q8_0

A pull that died mid-download leaves incomplete layers; re-running continues it, while a full disk or registry outage shows up here first.

Verify it worked

ollama run llama3.2

The model loads and you get a chat prompt. If it fails:

• Error: model “X” not found, try pulling it first — the name is wrong or not yet pulled; copy the exact string from the library and run ollama pull on it.
• Error: pull model manifest: file does not exist — a wrong repo or tag, or a bitsandbytes (BnB) repo Ollama cannot use; choose a GGUF repo and the exact quant tag.
• max retries exceeded / unexpected EOF — the network dropped; re-run the pull, and check free disk against how much RAM or VRAM a model needs.

Next: package a local file with a Modelfile, or install Ollama on Windows or macOS and Linux. GGUF-from-Hub details: Hugging Face’s Ollama guide.