Token Factory › Models
Model catalog
Every major model, one contract
Frontier, open-source, and Thai-native models behind one protocol. Switch with a model ID, not a migration.
Compatibility
Keep the SDK you already use
Point your existing client at Token Factory. The request style you send decides the response style you get back.
POST /v1/chat/completionsPOST /v1/messagesPOST /v1beta/models/*:generateContentCatalog
27 models across five modalities
Every model shares the same keys, allowlists, logs, and bill. SiamLLM is highlighted as our own model.
| Model ID | Provider | Best for |
|---|---|---|
siamllm-th | SiamGPT | Thai-native chat, Thai documents, local context |
anthropic/claude-fable-5 | Anthropic | Long context, complex reasoning |
anthropic/claude-sonnet-5 | Anthropic | Coding and agents at balanced cost |
anthropic/claude-opus-4.8 | Anthropic | Code and long-running agents |
openai/gpt-5.6-sol | OpenAI | General purpose, wide ecosystem |
openai/gpt-5.5 | OpenAI | General purpose |
openai/gpt-oss-120b | OpenAI | Open weights, private deployment |
google/gemini-3.1-pro-preview | Multimodal reasoning, science | |
google/gemini-3.5-flash | Fast, low-cost high volume | |
deepseek/deepseek-v4-pro | DeepSeek | Code, strong value |
qwen/qwen3.7-max | Alibaba Qwen | Multilingual, Chinese |
moonshotai/kimi-k2.7-code | Moonshot | Coding, long context |
z-ai/glm-5.2 | Z.ai | Enterprise applications |
meta-llama/llama-4-maverick | Meta | Open weights, fine-tuning base |
mistralai/mistral-medium-3-5 | Mistral | European open models |
| Model ID | Provider | Best for |
|---|---|---|
openai/gpt-image-2 | OpenAI | Photorealistic product and lifestyle images |
google/nano-banana-pro | High-detail image generation and editing | |
google/nano-banana-2 | Fast image generation, layouts |
| Model ID | Provider | Best for |
|---|---|---|
kling-v3-pro | Kuaishou Kling | Cinematic text-to-video |
kling-o3-pro | Kuaishou Kling | Multi-subject scenes, camera moves |
seedance-2-0 | ByteDance | Commercial product shots |
wan-2-7-t2v | Alibaba | Text-to-video, stylized scenes |
| Model ID | Provider | Best for |
|---|---|---|
fish-audio/tts | Fish Audio | Text-to-speech, voice cloning |
nvidia/riva-nmt | NVIDIA | Neural machine translation, Thai and English |
| Model ID | Provider | Best for |
|---|---|---|
nvidia/nemo-retriever-embedding | NVIDIA | Retrieval-augmented generation |
nvidia/nemo-retriever-reranking | NVIDIA | Second-pass ranking for search |
openai/text-embedding-3-small | OpenAI | General-purpose vectors, low cost |
Catalog shown for planning. Availability by delivery mode and region is confirmed at onboarding, and new models are added as they are validated.
Our own model
Need Thai to sound native?
SiamLLM is a large language model platform designed specifically for the Thai language, and it is available through the same API as every model above.
model: "siamllm-th" Meet SiamLLM
Missing a model you need?
Tell us which model and what volume. We onboard models based on customer demand.