OmniTranslate is a massively multilingual machine translation model supporting over 500 languages. Fine-tuned from Qwen 3 0.6B (with Unsloth), this model is designed for translation tasks on any device!

Features

500+ Languages Supported: The broadest coverage of languages supported for a translation model that's under 1 billion parameters!
Tiny Size: Beats any other large model on speed and memory usage. No other model is able to compete with this!

Issues

Accuracy on Common Languages: Accuracy on common languages in the dataset (like Spanish, Chinese, Romanian) is generally very good! Sometimes there's a chance that OmniTranslate can make hiccups. Examples are roșă and ami when translating to Romanian.
Accuracy on Rare Languages: Accuracy on rare languages in the dataset (like Toki Pona) isn't as good as on common languages!

As it follows, OmniTranslate 1.0 is a experimental model and shouldn't be used for tasks where accurate translations matter.

Notes

Providing the ISO code instead of the language name can improve the results a lot.

Usage

Code is by Gemini 3 Flash (then some little modifications by myself):

python

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

# 1. Load from your Hugging Face Repo
model_id = "MihaiPopa-1/OmniTranslate-1.0"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float32, # Standard for CPU
    device_map="cpu"           # Forces CPU usage
)

# 2. Translate (replace ron_Latn with your language here)
prompt = "<|im_start|>user\nTranslate to ron_Latn: OmniTranslate is a massively multilingual machine translation model supporting over 500 languages!<|im_end|>\n<|im_start|>assistant\n<think>\n"
inputs = tokenizer(prompt, return_tensors="pt").to("cpu")

with torch.no_grad():
    outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.1)
    
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Data Used

I used my own OmniSurgical 1.0, which the dataset itself is a extract of HF's FineTranslations.

120 sentences per language (60 per language pair).

Uploaded finetuned model

Developed by: MihaiPopa-1
License: apache-2.0
Finetuned from model : unsloth/qwen3-0.6b-unsloth-bnb-4bit

This qwen3 model was trained 2x faster with Unsloth and Huggingface's TRL library.