>_TheQuery
← Glossary

Doubao 1.5 Pro

LLM Models

ByteDance's reasoning model with Deep Thinking mode, matching GPT-4o performance at 50x lower cost with 256K context window.

Doubao 1.5 Pro, released by ByteDance in January 2025, is a large language model featuring a sparse Mixture-of-Experts (MoE) framework that activates only a subset of parameters during inference, delivering dense-model performance at a fraction of the computational cost. It supports context windows from 32,000 to 256,000 tokens.

The model matches GPT-4o in reasoning tasks and outperforms OpenAI's o1-preview and o1 on the AIME benchmark for mathematical reasoning. ByteDance claims it surpasses GPT-4o and Claude 3.5 Sonnet in knowledge retention, coding, reasoning, and Chinese language processing. The accompanying Doubao-1.5-vision-pro provides multimodal capabilities for image and video understanding.

Doubao 1.5 Pro's most remarkable feature is its cost efficiency - operational expenses are 5x lower than DeepSeek and over 200x lower than OpenAI's o1 model. The Deep Thinking mode enhances reasoning for complex problem-solving tasks. Doubao is the most popular AI app in China with over 80 million monthly active users, and ByteDance's AI business was a key driver of the company's $400 billion+ valuation.

Last updated: February 22, 2026