🇹🇭 Thai AI · Free to Use

ThaiLLM Complete Guide

Thai AI, Free to Use · NECTEC · VISTEC · AIEAT · KBTG · BDI

8 Chapters · ~60 Minutes · 86-Page PDF · Thai Language

National Thai AI developed by NECTEC · VISTEC · AIEAT · KBTG · BDI, trained on 100+ billion Thai tokens, processed 100% on Thailand's ThaiSC supercomputer. This guide walks you through setting up ThaiLLM in VSCode + Python. Free to use, no cost.

— Intro Video

Intro Video · ThaiLLM Tutorial in 60 Seconds

60 seconds · Learn to set up ThaiLLM in VSCode + Python · Thai audio

— Why ThaiLLM

Why Use Thai AI

Thailand is one of only four Asian countries with its own LLM (alongside China, Japan, and South Korea). All data stays within Thailand, never exported abroad.

  • 🇹🇭 Data stays in Thailand 100% — Processed on Thailand's ThaiSC supercomputer, no data sent abroad
  • 💎 100+ billion tokens — Trained on massive national Thai datasets, deep understanding of Thai context
  • 🆓 Free to use · Open API — Free API key, OpenAI-compatible, works with VSCode · Continue · Cline immediately
  • 🏛️ National partners — NECTEC · VISTEC · AIEAT · KBTG · BDI under the Ministry of Higher Education, Science, Research and Innovation
  • ⚡ 4 models to choose from — Pathumma (Think) · OpenThaiGPT · Typhoon (128K) · THaLLE (Finance)
  • 🌍 Built-in translation — Pathumma translates Thai ↔ EN, JP, CN, KR and 10+ more languages in just 3 lines of code
— 4 Models · Free to Use Now

Choose the Right Model for Your Task

Each model has unique strengths · One API key works with all four models

  • ⭐ Pathumma-ThaiLLM-qwen3-8b-think (by NECTEC · Thinking Mode) — AI thinks before answering · More accurate responses · Ideal for analytical tasks + translation
  • OpenThaiGPT-ThaiLLM-8B-Instruct (by AIEAT · Research Preview) — Fully open-source · General-purpose · Good Thai language understanding
  • Typhoon-S-ThaiLLM-8B-Instruct (by SCB 10X · 128K context) — Supports context up to 128,000 tokens · Ideal for long documents
  • THaLLE-0.2-ThaiLLM-8B-fa (by KBTG · Finance Expert) — Finance specialist · Accurate answers for economic and investment questions
— Tutorial · 8 Chapters

Complete Installation and Usage in 8 Chapters

From signing up for an API key to writing Python translation code · Approximately 60 minutes total

  • Chapter 1 · Intro · Thai AI Has Arrived! — Learn about ThaiLLM in 10 seconds · Why Thailand is one of four Asian countries with its own LLM
  • Chapter 2 · What is ThaiLLM? Who developed it? — National partners: NECTEC · VISTEC · AIEAT · KBTG · BDI under the Ministry of Higher Education, Science, Research and Innovation
  • Chapter 3 · Four Models Available Free Now — Compare Pathumma · OpenThaiGPT · Typhoon · THaLLE · Choose the right one for your task
  • Chapter 4 · Sign Up for a Free API Key at thaillm.or.th — Go to thaillm.or.th · Log in with Google/Email · Click Create API Key · Copy and save immediately (shown only once!)
  • Chapter 5 · Install Continue in VSCode — Ctrl+Shift+X · Search for "Continue" · Install · Or use existing Cline (OpenAI-compatible)
  • Chapter 6 · Configure config.yaml to Connect to ThaiLLM — Open config.yaml · Set apiBase: https://api.thaillm.or.th/v1 · apiKey: YOUR_KEY · Done
  • Chapter 7 · Demo · Translate with 3 Lines of Python — Write code to call Pathumma for instant Thai → EN, JP, CN, KR translation (see code below)
  • Chapter 8 · Outro · Start Using It Today — Summary of all key links · 100% free · No additional costs · See the complete 86-page guide in the PDF file
— Code Example

Translate with 3 Lines of Python

Use the openai SDK, point base_url to the ThaiLLM API · Done

# pip install openai
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_THAILLM_API_KEY",
    base_url="https://api.thaillm.or.th/v1",
)

def translate(text, target_lang="English"):
    resp = client.chat.completions.create(
        model="Pathumma-ThaiLLM-qwen3-8b-think-3.0.0",
        messages=[{"role": "user",
                    "content": f"Translate to {target_lang}: {text}"}],
        temperature=0.3,
     )
    return resp.choices[0].message.content

print(translate("Hello, welcome", "Japanese"))
# → こんにちは、ようこそ
// DOWNLOAD

Complete PDF Guide · 86 Pages

Every step · Every code snippet · Every example, plus a Streamlit translation web app to finish · 📕 86 pages · 📦 1.85 MB · ✅ Free, no ads