Open-Source Model Powers Structural Biology Agent Demo
MaziyarPanahi · x · 2026-07-11
This post demonstrates a fully open-source structural biology agent workflow:
- Brain: GLM-5.2
- Vision: Qwen3-VL running via Llama.cpp
- Interface/Rendering: Open-source Mol viewer
- Orchestration: Pydantic AI
- Target: EGFR + erlotinib structure from RCSB PDB
The author mentions that the entire process used just a single prompt to generate 6 render versions. At each step, the vision model provides actual critiques, iterating until the drug's position in the pocket "looks more reasonable." Follow-up replies emphasize that this is a demonstration of open models capable of "doing structural biology on their own."
More from coding & agent
- Frontier models need ways to verify success — or they'll invent their own — daniel_mac8 · 2026-09-11
- Goal-driven AI needs verifiable success signals, or it invents its own — daniel_mac8 · 2026-09-11
- Is inference latency becoming the biggest bottleneck for production AI agents? — Euphoric_Sea632 · 2026-09-11
- Anthropic researcher: 99% of engineers now run swarms of 300+ self-improving agents — AlishaOutridge · 2026-09-11
- Gergely Orosz: Shipping 10x PRs With AI Agents, Sites Fill With Small Regressions — ducha_aiki · 2026-09-11
- Same Echo Maze prompt, three frontier models: all passed visually but shipped the same hidden bug — eyishazyer · 2026-09-11