iapp/openthai2.0-legal-thaillm-nemotron-3-nano-30b-a3b
Text Generation • 32B • Updated • 145 • 3
The models to use today. Start here.
Note ⚖️ Thai law, statute citation, legal RAG. 30B MoE, ~3B active — runs on one 24 GB GPU.
Note 🧠 Reasoning, maths, logic, code in Thai. Beats 70B models at half the size.
Note 💬 General Thai chat and coding. 98.2% Thai language accuracy.
Note 💻 Smallest option — for limited GPU budgets.