Falcon-H1: A Family of Hybrid-Head Language Models Redefining Efficiency and Performance

📚 Paper | 💻 GitHub Repository | 📄 Project Documentation | 🖥️ Hugging Face Demo

Table of Contents

0. TL;DR
1. Model Details
2. Training Details
3. Usage
4. Evaluation
5. Citation

TL;DR

Falcon-H1 is a new series of large language models (LLMs) featuring a novel hybrid architecture that combines Transformer-based attention with State Space Models (SSMs) for superior long-context memory and computational efficiency. Released in multiple configurations (0.5B to 34B parameters), Falcon-H1 models demonstrate state-of-the-art performance and exceptional parameter and training efficiency. The flagship Falcon-H1-34B matches or outperforms models up to 70B scale, while smaller models also show strong performance relative to their size. These models excel across reasoning, mathematics, multilingual tasks, instruction following, and scientific knowledge, supporting up to 256K context tokens and 18 languages. All models are released under a permissive open-source license.

Model Details

Model Description

• Developed by: https://www.tii.ae
• Model type: Causal decoder-only
• Architecture: Hybrid Transformers + Mamba architecture
• Language(s) (NLP): English, Multilingual
• License: Falcon-LLM License

Training details

For more details about the training protocol of this model, please refer to the Falcon-H1 technical blogpost and Technical Report.

Usage

Currently to use this model you can either rely on Hugging Face transformers, vLLM or our custom fork of llama.cpp library.

Inference

Make sure to install the latest version of transformers or vllm, eventually install these packages from source:

pip install git+https://github.com/huggingface/transformers.git

Refer to the official vLLM documentation for more details on building vLLM from source.

🤗 transformers

Refer to the snippet below to run H1 models using 🤗 transformers:

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "tiiuae/Falcon-H1-1B-Base"

model = AutoModelForCausalLM.from_pretrained(
  model_id,
  torch_dtype=torch.bfloat16,
  device_map="auto"
)

# Perform text generation

vLLM

For vLLM, simply start a server by executing the command below:

# pip install vllm
vllm serve tiiuae/Falcon-H1-1B-Instruct --tensor-parallel-size 2 --data-parallel-size 1

🦙 llama.cpp

Falcon-H1 is compatible with the newest version of llama.cpp

Evaluation

Falcon-H1 series perform very well on a variety of tasks, including reasoning tasks.

You can check more in detail on our our release blogpost, detailed benchmarks.

Useful links

• View our release blogpost.
• View our technical report.
• Feel free to join our discord server if you have any questions or to interact with our researchers and developers.

Citation

If the Falcon-H1 family of models were helpful to your work, feel free to give us a cite.

@article{falconh1,
    title={Falcon-H1: A Family of Hybrid-Head Language Models Redefining Efficiency and Performance},
    author={Jingwei Zuo and Maksim Velikanov and Ilyas Chahed and Younes Belkada and Dhia Eddine Rhayem and Guillaume Kunsch and Hakim Hacid and Hamza Yous and Brahim Farhat and Ibrahim Khadraoui and Mugariya Farooq and Giulia Campesan and Ruxandra Cojocaru and Yasser Djilali and Shi Hu and Iheb Chaabane and Puneesh Khanna and Mohamed El Amine Seddik and Ngoc Dung Huynh and Phuc Le Khac and Leen AlQadi and Billel Mokeddem and Mohamed Chami and Abdalgader Abubaker and Mikhail Lubinets and Kacper Piskorski and Slim Frikha},
    journal = {arXiv preprint arXiv:2507.22448},
    year={2025}
}