Launched with important enhancements, Alibaba’s Qwen-3 Max is able to compete with AI giants like OpenAI’s GPT-5, Google’s Gemini 2.5 Professional, and Anthropic’s Claude Opus 4. Be taught Extra.

Alibaba has launched its strongest AI but on September 24, 2025. In response to the Alibaba Cloud web site, the open-source Qwen collection noticed 7 million downloads by Might 2024 (simply after launching in April 2023). Notably, Qwen is now ranked third globally on the LMArena textual content leaderboard. To greatest serve its customers, this Chinese language AI has made important enhancements to its new model. So, the brand new mannequin is all set to compete with AI giants like OpenAI’s GPT-5, Google’s Gemini 2.5 Professional, and Anthropic’s Claude Opus 4. How is the brand new model totally different? How does it rank in comparison with others? Be taught extra. 

What Is Qwen-3 Max?

• Qwen-3 Max is a part of Alibaba’s Qwen collection of AI fashions, often known as the Qwen household.
• That is the primary Qwen mannequin to surpass the 1 trillion parameters (these parameters are the connections within the mind of the AI. Extra parameters imply a extra superior AI mannequin).
• Qwen-3 Max was educated on a whopping 36 trillion tokens (these tokens are phrases that an AI reads whereas getting educated).
• The capabilities of this mannequin are such that it could learn, keep in mind and course of large quantities of textual content unexpectedly (as an example, a complete ebook, lengthy analysis paper, and even full codebases).
• In technical phrases, it could take a context size of as much as 1 million tokens.

What Can Qwen-3 Max Do Higher Now?

In response to Alibaba, the Qwen-3 Max is improved in:

• Reasoning (that means figuring issues out step-by-step).
• Instruction following (the mannequin can perceive and perform all of the duties you give it).
• Multilingual assist (for now, the mannequin can work nicely in each English and Chinese language).
• Lengthy-tail data protection (that means the AI now is aware of about extra advanced topics which can be much less widespread).
• Math, coding, logic, and science (this model is now educated to unravel any technical duties extra precisely).
• Diminished hallucinations (this model is much less prone to make issues up as in comparison with its older fashions).
• Excessive-quality response to open-ended or advanced questions.

How Does It Rank In comparison with Others?

Lmarena’s Textual content Leaderboard

• Qwen-3 Max is ranked third on the LMArena’s textual content leaderboard (it is a rating given to AI text-based fashions.
• It’s forward of GPT-5.
• And a bit of behind Gemini 2.5 Professional, Claude Opus 4.1 (pondering mode), OpenAI’s o3, 4-o, GPT-4.5, and GPT-5 Excessive fashions.

SWE-Bench Verified 

• Qwen-3 Max has scored 69.6. on SWE-Bench Verified (it’s a take a look at for AI fashions to rank their actual coding problem-solving talent).
• It has scored higher than DeepSeek V3.1 and just a bit under Claude Opus 4.

Tau2-Bench

• Qwen-3 Max has scored 74.8 on the Tau2-Bench (it is a take a look at for AI to see how nicely the makes use of instruments in an agent-like means).
• On this take a look at, Qwen-3 is nicely forward of Claude Opus 4 and DeepSeek V3.1.

How Can You Use Qwen-3 Max?

• Qwen is an open-source AI mannequin. The brand new Qwen-3 Max can be free.
• One can entry the brand new model on the Qwen app (accessible on iOS and Android) or straight on the Qwen web site.
• In case you are a brand new consumer, the app will default to put in as Qwen-3 Max.
• In case you do not see the brand new model, here is what you are able to do:
• Faucet on the highest left nook of the Qwen app after which choose the Qwen-3 Max mannequin, and there you go.
• It is similar to ChatGPT or Gemini, you may work together and ask the AI any query you’ve.

WIDGET: questionnaire | CAMPAIGN: Easy Questionnaire