Shapelearn Qwen 3.8 27B (13.1 GB VRAM)
Hacker News users discussed the new Shapelearn Qwen 3.8 AI model, which has 27 billion parameters and requires about 13.1 GB of VRAM. The thread noted its size, hardware needs, and sparked interest in evaluating its text‑generation and code‑completion abilities, while debating the balance between capability and memory usage for various downstream applications in research and industry.