HomeQwen2.5-72B-Instruct
72B
Model parameter quantity
Alibaba
Affiliated organization
Open Source
License Type
September 19, 2024
Release time
Related figures
Model Introduction
Like Qwen2, the Qwen2.5 language models support up to 128K tokens and can generate up to 8K tokens. They also maintain multilingual support for over 29 languages, including Chinese, English, French, Spanish, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, Arabic, and more.
Comprehensive score
Language dialogue
Knowledge reserve
Reasoning association
Mathematical calculation
Code writing
Command following
Swipe left and right to view more
Language comprehension ability
Often makes semantic misjudgments, leading to obvious logical disconnects in responses.
5.9
Knowledge coverage scope
Possesses core knowledge of mainstream disciplines, but has limited coverage of cutting-edge interdisciplinary fields.
8.5
Reasoning ability
Unable to maintain coherent reasoning chains, often causing inverted causality or miscalculations.
6.5
Model comparison
Qwen2.5-72B-Instruct vs Doubao-Seed-1.6-251015 (Thinking)
The deep reasoning model released by ByteDance, which supports manual switching of deep reasoning, and its performance is significantly improved compared to doubao-1.5.
Qwen2.5-72B-Instruct vs Doubao-Seed-1.6-thinking-250715
The latest version of the seed series model launched by ByteDance, which supports the thinking mode.
Related model
Qwen3.6-27B
Qwen3.6 is the next-generation open-source model in the Qwen family and an upgraded version of the Qwen3.5 series.
Qwen3.5-397B-A17B
Qwen3.5 is a significant foundation model recently launched by Alibaba, featuring comprehensive upgrades in multimodal fusion, efficient hybrid architecture, scalable reinforcement learning, multilingual coverage and training infrastructure, with greatly improved performance and practicality.
Qwen3.6-Max-Preview
Qwen3.6-Max-Preview is Qwen’s flagship proprietary model, focusing on stronger reasoning, complex task handling, and high-quality conversational performance.
Qwen3-Next-80B-A3B-Thinking
The latest released Qwen3-Next series in Qwen models, improving scaling efficiency through innovative model architecture.
Qwen3-235B-A22B-Thinking-2507
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
Relevant documents
Meituan CEO Wang Xing: AI Agents Hit Harder Than ChatGPT
During a management communication session on March 13, 2026, Meituan CEO Zhang Wang discussed the trajectory of artificial intelligence (AI). He argued that AI-driven transformation will surpass the impact of the entire Internet era. Drawing a metaph
Jensen Huang Reveals Nvidia’s ‘Brand New’ $200B Market Opportunity
Nvidia founder and CEO Jensen Huang stands out as one of the most compelling corporate visionaries, if not the greatest, in driving his company’s narrative. He may even eclipse Salesforce’s Marc Benioff in his unwavering optimism regarding the firm’s
Hugging Face in Talks for $13B Acquisition
Business Insider reported over the weekend that Hugging Face has been approached to sell at a valuation of $13 billion or more.The startup’s platform and open-source community is where developers and researchers share, find, test, and deploy AI model
Lovable Launches Mobile AI Coding App with Voice Prompt Support for Web App Generation
On April 28, Lovable officially launched its no-code AI app builder for iOS and Android. Centered on the "Vibe Coding" concept, the platform lets users capture ideas via voice or text prompts and instantly develop apps on mobile devices. Users can se
Superintelligence Arrives: Do We Let It In?
Loading the player…AI companies have been talking about superintelligent AI like it’s inevitable, but recent safety incidents like OpenAI’s Hugging Face breach are demonstrating the potential dangers of deploying AI systems that are more capable than
