Youtu-LLM

🎯 Brief Introduction

Youtu-LLM is a new, small, yet powerful LLM, contains only 1.96B parameters, supports 128k long context, and has native agentic talents. On general evaluations, Youtu-LLM significantly outperforms SOTA LLMs of similar size in terms of Commonsense, STEM, Coding and Long Context capabilities; in agent-related testing, Youtu-LLM surpasses larger-sized leaders and is truly capable of completing multiple end2end agent tasks.

Youtu-LLM has the following features:

• Type: Autoregressive Causal Language Models with Dense MLA
• Release versions: Base and Instruct
• Number of Parameters: 1.96B
• Number of Layers: 32
• Number of Attention Heads (MLA): 16 for Q/K/V
• MLA Rank: 1,536 for Q, 512 for K/V
• MLA Dim: 128 for QK Nope, 64 for QK Rope, and 128 for V
• Context Length: 131,072
• Vocabulary Size: 128,256

🤗 Model Download

📰 News

• [2026.01.28] You can now directly use Youtu-LLM with Transformers>=5.1.0.
• [2026.01.07] You can now fine-tune Youtu-LLM with ModelScope.
• [2026.01.04] You can now fine-tune Youtu-LLM with LlamaFactory.

📊 Performance Comparisons

Instruct Model

General Benchmarks

Agentic Benchmarks

🚀 Quick Start

This guide will help you quickly deploy and invoke the Youtu-LLM-2B model. This model supports "Reasoning Mode", enabling it to generate higher-quality responses through Chain of Thought (CoT).

If you wish to use Youtu-LLM-2B based on earlier versions of transformers, please make sure to download the model repository before this commit.

1. Environment Preparation

Ensure your Python environment has the transformers library installed and that the version meets the requirements.

pip install "transformers>=4.56.0,<=4.57.1" torch accelerate

Note

• (1) We recommend to limit the version of transformers: pip install "transformers>=4.56.0,<=4.57.1", which is comparable with the current remote codes;
• (2) Do not use transformers==4.57.2, since there is a bug unfixed;
• (3) If you would like to maintain a higher version (e.g., 4.57.3), you should slightly modify the "check_model_inputs" in modeling_youtu.py to "check_model_inputs()", following the patch.

2. Core Code Example

The following example demonstrates how to load the model, enable Reasoning Mode, and use the re module to parse the "Thought Process" and the "Final Answer" from the output.

import re
from transformers import AutoTokenizer, AutoModelForCausalLM

# 1. Configure Model
model_id = "tencent/Youtu-LLM-2B"

# 2. Initialize Tokenizer and Model
tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    trust_remote_code=True
)

# 3. Construct Dialogue Input
prompt = "Hello"
messages = [{"role": "user", "content": prompt}]

# Use apply_chat_template to construct input; set enable_thinking=True to activate Reasoning Mode
input_text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True
)

model