TeichAI/Qwen3-4B-Instruct-2507-Polaris-Alpha-Distill

Qwen3 4B Instruct 2507 x Polaris Alpha

This is a non-reasoning model trained on 1,000 examples from Polaris Alpha, an early snapshot of GPT-5.1 with reasoning effort set to minimal.

GGUFs available here

---

• Developed by: TeichAI
• License: apache-2.0
• Finetuned from model : unsloth/Qwen3-4B-Instruct-2507

This qwen3 model was trained 2x faster with Unsloth and Huggingface's TRL library.