Dicta-LM 3.0 is a powerful open-weight collection of LLMs, trained on extensive corpora of Hebrew and English texts. The models are available for download and for unlimited use. The models set a new SOTA for their weight-class for Hebrew, both as base models and chat models.
This is the 12-billion-parameter Hybrid-SSM instruct model, originally initialized from
NVIDIA-Nemotron-Nano-12B-v2-Base.
This version of the model is dynamically quantized to FP8, utilizing the Hopper and Blackwell architectures for faster inference with a lower memory footprint.
For full details of this model please read our
release blog post or the
technical report.
You can view and access the full collection of base/instruct unquantized/quantized versions of
DictaLM 3.0 here.
In order to leverage instruction fine-tuning, your prompt should be rendered using the chat template specified for this model. Most libraries deal with this automatically, so you can just let them do it.
1from openai import OpenAI
2
3client = OpenAI(
4 base_url="http://localhost:8000/v1",
5 api_key="sk-no-key-required"
6)
7
8response = client.chat.completions.create(
9 model="dicta-il/DictaLM-3.0-Nemotron-12B-Instruct-FP8",
10 messages=[
11 {"role": "user", "content": "Hello, how are you?"}
12 ],
13)
14
15print(response.choices[0].message.content)
The model supports tool-calling, enabling integration with external tools and APIs. For example how to use the tool calling, see the
vLLM documentation.
1@article{Shmidman2025DictaLM3,
2 title={{Dicta-LM 3.0: Advancing The Frontier of Hebrew Sovereign LLMs}},
3 author={Shaltiel Shmidman and Avi Shmidman and Amir DN Cohen and Moshe Koppel},
4 year={2025},
5 publisher={{DICTA / Jerusalem, Israel}},
6 note={https://www.dicta.org.il/publications/DictaLM_3_0___Techincal_Report.pdf}
7}
