openthai/ openthai-1.6-72b:q4_K_M

1 Download Updated 7 hours ago

OpenThai's general-purpose flagship — 98.2% Thai language accuracy, strongest of the family at Thai coding. ~48 GB RAM/VRAM.

ollama run openthai/openthai-1.6-72b:q4_K_M

curl http://localhost:11434/api/chat \
  -d '{
    "model": "openthai/openthai-1.6-72b:q4_K_M",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='openthai/openthai-1.6-72b:q4_K_M',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'openthai/openthai-1.6-72b:q4_K_M',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 7 hours ago

7 hours ago

896928651dbe · 47GB ·

model

archqwen2

·

parameters72.7B

·

quantizationQ4_K_M

47GB

template

{{ if .System }}<|im_start|>system {{ .System }}<|im_end|> {{ end }}{{ if .Prompt }}<|im_start|>user

182B

system

You are OpenThai, a Thai language assistant developed by the OpenThai project (AIEAT and iApp Techno

225B

license

See https://huggingface.co/openthaigpt — licence metadata is other/qwen. OpenThai (formerly OpenTh

199B

params

{ "num_ctx": 32768, "stop": [ "<|im_start|>", "<|im_end|>" ], "tempe

105B

Readme

OpenThai 1.6 72B 🇹🇭

The general-purpose flagship — 98.2% Thai language accuracy, strongest of the family on Thai coding.

ollama run openthai/openthai-1.6-72b

Balanced across mathematics, coding and Thai language understanding, with a large coding jump over 1.5 (LiveCodeBench-TH 32.43 vs 12.61) and the family’s best general average.

Benchmarks (one identical protocol)

Benchmark	OTG 1.6 72B	Typhoon2 L3.1 70B	Pathumma 7B
LiveCodeBench-TH	32.43	27.03	0
MATH500	82.00	67.40	42.80
OpenThaiEval	78.70	72.54	65.27
Thai language accuracy	98.20	99.80	98.60
Average (8 tasks)	52.34	47.91	28.56

Full table: Thai LLM Leaderboard · Technical report (arXiv:2504.01789)

Requirements & tips

~48 GB RAM/VRAM (Q4_K_M) — a 64 GB Mac or 2× 24 GB GPUs
Limited hardware? openthai-1.5-7b runs in 8 GB
Want reasoning? openthai-r1-32b is smaller and better at it
OpenAI-compatible API at http://localhost:11434/v1

Links

Website · GGUF weights (Q4/Q5/Q8) · Discord

Free and open-weight, from the OpenThai project (formerly OpenThaiGPT) — AIEAT × iApp Technology.

# OpenThai 1.6 72B 🇹🇭

**The general-purpose flagship — 98.2% Thai language accuracy, strongest of the family on Thai coding.**

```
ollama run openthai/openthai-1.6-72b
```

Balanced across mathematics, coding and Thai language understanding, with a large coding
jump over 1.5 (LiveCodeBench-TH 32.43 vs 12.61) and the family's best general average.

## Benchmarks (one identical protocol)

| Benchmark | **OTG 1.6 72B** | Typhoon2 L3.1 70B | Pathumma 7B |
|---|---|---|---|
| LiveCodeBench-TH | **32.43** | 27.03 | 0 |
| MATH500 | **82.00** | 67.40 | 42.80 |
| OpenThaiEval | **78.70** | 72.54 | 65.27 |
| Thai language accuracy | 98.20 | **99.80** | 98.60 |
| **Average (8 tasks)** | **52.34** | 47.91 | 28.56 |

Full table: [Thai LLM Leaderboard](https://huggingface.co/spaces/openthaigpt/thai-llm-leaderboard) ·
[Technical report (arXiv:2504.01789)](https://arxiv.org/abs/2504.01789)

## Requirements & tips

- **~48 GB RAM/VRAM** (Q4_K_M) — a 64 GB Mac or 2× 24 GB GPUs
- Limited hardware? [openthai-1.5-7b](https://ollama.com/openthai/openthai-1.5-7b) runs in 8 GB
- Want reasoning? [openthai-r1-32b](https://ollama.com/openthai/openthai-r1-32b) is smaller *and* better at it
- OpenAI-compatible API at `http://localhost:11434/v1`

## Links

[Website](https://openthai.aieat.or.th/ollama) ·
[GGUF weights (Q4/Q5/Q8)](https://huggingface.co/openthaigpt/openthaigpt-1.6-72b-instruct-GGUF) ·
[Discord](https://discord.gg/7KDdKkBGUs)

*Free and open-weight, from the OpenThai project (formerly OpenThaiGPT) — AIEAT × iApp Technology.*

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)