Tech Desk 
Nov 28, 2024
As the present era is witnessing fierce competition among tech giants willing to surpass one another with innovative and powerful artificial intelligence (AI) tools, Alibaba has released a new AI model named QwQ-32B-Preview.
Developed by the Chinese e-commerce juggernaut's Qwen team, the tool is said to stand tall against OpenAI's ChatGPT o1, as it’s the first AI model available to download with a permissive license.
The most significant aspect of the AI model is that it's capable of comprehending prompts of up to 32,000 words in length since it has 32.5 billion parameters.
Alibaba new AI model
Compared to its direct archrival ChatGPT o1-preview and o1-mini, QwQ-32B-Preview has performed exceptionally better on certain benchmarks, proving to be a formidable opponent for OpenAI's two reasoning models.
It should be noted in the context of AI that parameters refer to a model’s problem-solving skills, meaning AI models with a greater number of parameters usually perform better than those with fewer parameters.
Since OpenAI has so far not revealed the parameter count for its reasoning models, the benchmark threshold of the QwQ-32B-Preview proves that they possess fewer parameters than those embedded under Alibaba's AI model.
According to testing conducted by Alibaba itself, QwQ-32B-Preview beats OpenAI’s o1 models on the AIME and MATH tests, where the former uses third-party models to assess a model’s performance, whereas the latter, MATH, collectively represent word problems.
Alibaba's QwQ-32B-Preview is capable of solving logic puzzles and reasonably challenging math problems with its “reasoning” capabilities.
But there seems to be a margin of downsides, as it can switch languages unexpectedly, get stuck in loops, and misunderstand prompts requiring “common sense reasoning,” stated a blog post on the official Alibaba website.
