OpenThaiGPT
Open-source Thai-centric large language model developed by Thailand's AI research associations under the national ThaiLLM effort.
Overview
OpenThaiGPT is an open-source Thai-centric large language model developed by Thai research groups including the AI Entrepreneur Association of Thailand (AIEAT), the AI Association of Thailand (AIAT) and NECTEC. Released under the Apache 2.0 licence, later versions such as OpenThaiGPT 1.5 build on Qwen 2.5 and are fine-tuned on over two million Thai instruction pairs, with reasoning variants added in 1.6 and R1. It offers models across a range of sizes and reports state-of-the-art results among open Thai models.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 24 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
ASEAN Perspective
OpenThaiGPT in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
What this is for: A Thai-centric open LLM for chat and instruction-following tuned specifically for the Thai language.
Who this is for: Thai developers, businesses and researchers wanting a permissively licensed local-language model they can self-host and adapt.
Availability: Free and open-source under Apache 2.0, with weights on Hugging Face; self-hosted deployment.
What people say
OpenThaiGPT has no consumer-review presence, but among open Thai models it has some of the strongest documented standing. Released under Apache 2.0 with weights on Hugging Face across 7B, 14B and 72B sizes, its 1.5 release is backed by an arXiv paper and is frequently cited as achieving the highest average scores among open-source Thai LLMs on Thai-language exams.
The benchmark detail is worth stating precisely: on the Thai Exam benchmark, the 72B model reportedly scored about 64.07% — competitive with GPT-4o (63.89%) and Llama 3.1 405B (63.54%), while sitting below Claude 3.5 Sonnet (69.2%). Being permissively licensed, self-hostable and backed by Thailand's AI associations (AIEAT, AIAT) and NECTEC gives it real credibility with Thai developers who want a local model they can adapt.
The caveats: those results come largely from the project's own paper and benchmark, so independent verification is limited, and matching much larger closed models on one Thai exam suite doesn't mean parity on general reasoning or coding. It's built on Qwen rather than trained from scratch, and as base/instruct open weights it needs infrastructure and tuning to deploy — there's no polished managed product for non-technical users.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including OpenThaiGPT's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with OpenThaiGPT unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to OpenThaiGPT directly →
Spotted something out of date? Suggest an update →
Alternatives to OpenThaiGPT
More in LLMs & Chat
