Typhoon Documentation & Developer Guide

👋 Welcome to Typhoon Documentation

Explore Typhoon — Thai-first, multilingual large language models optimized for practical deployment and developer accessibility. Whether you’re prototyping in a notebook, scaling an AI product, or just getting started with automation, you’ll find everything you need here.

🗺️ Overview: Where & How to Access Typhoon Models

Typhoon is available across multiple platforms. Choose what fits your workflow and infrastructure:

0. Web Playground 🌐

Best for: Fast experimentation and prompt design

No setup required — just start typing. Our playgrounds support both text and multimodal input.

• 🔗 Typhoon 2 & 2.1 Playground

• 🔗 Typhoon T1 (Thai reasoning) – Research Preview

• 🔗 Typhoon Audio - Research Preview

• 🔗 Typhoon Vision - Research Preview

• 🔗 Typhoon OCR - Research Preview

1. API-based Access 🔌

Best for: Web apps, agents, automation workflows, and scalable integrations

Includes everything where you send a request and get a response over HTTP — regardless of where the model is hosted. This would include:

Typhoon API (hosted by Typhoon team)

Together.ai (official partner)

OpenRouter (community router)

Cloud Marketplaces (e.g. AWS, Azure)

1.1 Official APIs

Typhoon API ✅ (Free Tier)

Key Features

• OpenAI-Compatible API: Drop-in replacement for existing OpenAI-based apps
• Streaming Support: Token-by-token real-time responses
• Zero cost for light usage — ideal for startups and prototyping
• Hosted and maintained by Typhoon

Available Models

Model ID	Size	Description	Context Window	Rate Limits	Release Date
typhoon-v2.1-12b-instruct	12B	Best model for instruction-following, tool use, and Thai NLP	56K tokens	5 req/s, 200 req/m	2025-05-05

We also have API for OCR model. Please see OCR tab for more information.

📘 API Documentation

Typhoon API Pro via Together.ai

If you’re looking to scale Typhoon in your business or product with guaranteed availability, high throughput, and simple billing, Together.ai is our official cloud partner.

✅ Why Together.ai?

• Production-ready infrastructure — optimized for low latency and stability

• Almost no rate limits — deploy large-scale agents or multi-user apps

• Simple pay-as-you-go billing — scale from startup to enterprise

🔧 Available Typhoon Models on Together.ai

Model	Size	Description	Endpoint	Note
Typhoon 2.1 Gemma	12B	Lightweight and powerful for most common use cases	🔗 Endpoint	.

📘 Read our quickstart guide

1.2 Third-party Routers and Marketplaces

OpenRouter 🔁

OpenRouter is a rising favorite among developers for its flexibility and ease of use. Typhoon models are now available on OpenRouter, making it easy to:

• 🔄 Compare Typhoon with other top open-source models
• ⚙️ Plug directly into automation tools like n8n, Zapier, and Make.com
• 🧩 Use Typhoon in custom chatbot frameworks and no-code/low-code agents. This is ideal for Thai startups building agents, lead-generation bots, or content automation flows.

📘 View Typhoon on OpenRouter

Cloud Marketplace Access ☁️

Deploy Typhoon in your existing cloud environment.

• Azure AI Model Catalog: Typhoon 1.5 available (Typhoon 2 models coming soon)

• AWS Marketplace: Coming soon

Please refer to respective platforms for deployment and billing setup.

2. Run Locally 💻

These tools support local use:

Use Typhoon entirely offline — great for R&D, privacy-sensitive use, or enterprise IT environments.

🤗 Hugging Face

• Fine-tune and run Typhoon models with the transformers library

• Use in offline, on-premise settings

• Access models in GGUF, Safetensors, and standard formats

Model Family	Sizes	Use Case
Typhoon2.1-Text	4B, 12B	Latest instruction & multilingual model
Typhoon2-Text	1B–70B	Versatile Thai-first models
Typhoon2-R1	70B	Advanced reasoning with SFT + DeepSeek merge
Typhoon T1	3B	Thai reasoning model, research preview
Typhoon2-Audio	8B	End-to-end speech-to-speech
Typhoon2-Vision	7B	Vision-Language model for image/video input

🔗 Explore Typhoon on Hugging Face

LM Studio & Ollama

Use Typhoon on your laptop with local tools. No GPU? No problem — use quantized models with CPU inference.

• LM Studio: GUI app for offline use (GGUF format)

• Ollama: Run with one CLI command

🔗 Typhoon on Ollama

📘 Ollama Quickstart Guide

📘 LM Studio Quickstart Guide

🧪 Get Started with Typhoon API

Jump in and start building with Thai-first LLMs in just a few steps:

1. 🔑 Create your API key — Sign up on the Typhoon Playground to generate your free API key.

2. 🚀 Follow the Quickstart Guide — Learn how to make your first API call with step-by-step instructions.

3. 💡 Explore real-world use cases + source codes! — See you can use Typhoon in business, automation, and research.

4. 💻 Browse example code and templates — Speed up your integration with ready-to-use API calls and workflows.

🧑‍🤝‍🧑 Support and Community

• 📧 Email: contact@opentyphoon.ai

• 💬 Discord: Join the Typhoon Community

• 📄 Papers: Typhoon | Typhoon 2 | Typhoon T1 | Typhoon R1