yairpatch

/

Qwen3-VL-30B-A3B-Instruct-GGUF

like 26

Follow

yair patch 15

Issue Summary

Getting a persistent 500 Internal Server Error: unable to load model when attempting to run the Q5_K_M quantization of this model with Ollama.

Environment

• Ollama Version: 0.12.5
• OS: Windows 10/11 (Build 26100)
• Model: hf.co/yairpatch/Qwen3-VL-30B-A3B-Instruct-GGUF:Q5_K_M
• Model Size: 22 GB (listed), 20.23 GB (blob file)

Steps to Reproduce

1. Pull the model: ollama pull hf.co/yairpatch/Qwen3-VL-30B-A3B-Instruct-GGUF:Q5_K_M
2. Model downloads successfully (all blobs at 100%)
3. Run: ollama run hf.co/yairpatch/Qwen3-VL-30B-A3B-Instruct-GGUF:Q5_K_M

Expected Behavior

Model should load and start an interactive chat session.

Actual Behavior

Error: 500 Internal Server Error: unable to load model: D:\OllamaModels\blobs\sha256-a3dcf99539e09f8a9f5578508bc0b834f62b0bd85e4764d56e942a9d89def85b

Troubleshooting Attempted

• ✅ Verified blob file exists and is the correct size (20.23 GB)
• ✅ Removed and re-downloaded the model (checksums verified)
• ✅ Restarted Ollama service
• ✅ Confirmed model appears in ollama list

Questions

1. Is the Q5_K_M quantization compatible with Ollama 0.12.5?
2. Are there known issues with Qwen3-VL vision models in Ollama?
3. What are the minimum system requirements (RAM/VRAM) for this quantization?
4. Would a different quantization (Q4_K_M, Q6_K, etc.) work better with Ollama?

Any guidance would be appreciated!

ps- ty cursor for writing this