Qwen 2.5 Max: AI battle heats up as Alibaba claims its new model outperforms Meta and DeepSeek in benchmarks

Qwen 2.5 Max: Alibaba claims its new AI model demonstrates superior performance over competitors such as Meta's Llama and DeepSeek’s V3, highlighting the competitive pressure from DeepSeek's rapid rise in the AI market.

Livemint, Written By Govind Choudhary, Jocelyn Fernandes
Updated30 Jan 2025, 02:25 PM IST
Qwen 2.5: Alibaba Group Holding has unveiled new benchmark scores for its latest artificial intelligence model, Qwen 2.5 Max, positioning it as a top contender in the global AI race.
Qwen 2.5: Alibaba Group Holding has unveiled new benchmark scores for its latest artificial intelligence model, Qwen 2.5 Max, positioning it as a top contender in the global AI race.(qwenlm.github.io)

Qwen 2.5: Alibaba Group on Wednesday unveiled new benchmark scores for its latest artificial intelligence model, Qwen 2.5 Max, positioning it as a top contender in the global AI race. 

According to figures shared by Alibaba Cloud via WeChat, the upgraded model outperformed Meta Platforms Inc’s Llama and DeepSeek’s V3 model in a variety of tests.

This development underscores Alibaba’s increasing investment in AI and cloud computing, where it competes fiercely with fellow Chinese tech giants such as Tencent and Baidu. All three firms are vying for dominance in China’s AI landscape, particularly in attracting developers to their platforms.

Also Read | How DeepSeek AI sent $593bn shockwave in tech stocks, rattled global markets

Qwen 2.5 Max vs DeepSeek AI: Alibaba Claims Dominance over ChatGPT, DeepSeek V3

A notable reference in Alibaba’s comparison is DeepSeek, a 20-month-old startup from Hangzhou, the same city where Alibaba is headquartered.

DeepSeek has rapidly gained global recognition and, according to Alibaba’s data, now serves as a primary benchmark for its AI advancements. The cloud computing arm of the e-commerce behemoth further suggested that Qwen 2.5 Max surpasses models from OpenAI and Anthropic in specific evaluation metrics.

The official account of Qwen posted on X, “The burst of DeepSeek V3 has attracted attention from the whole AI community to large-scale MoE models. Concurrently, we have been building Qwen2.5-Max, a large MoE LLM pretrained on massive data and post-trained with curated SFT and RLHF recipes. It achieves competitive performance against the top-tier models, and outcompetes DeepSeek V3 in benchmarks like Arena Hard, LiveBench, LiveCodeBench, GPQA-Diamond [sic].”

Also Read | Qwen 2.5 vs DeepSeek vs ChatGPT: The new battleground for AI supremacy
Also Read | Human rights groups slam China’s DeepSeek over data privacy and state propaganda

AI Sector Battle Toughens Up

China’s AI sector is heating up as companies aggressively compete for market share. Cloud service providers, including Alibaba and Tencent, have recently reduced their prices to attract users, with DeepSeek emerging as a significant factor in this intensifying price war. Several AI startups in China have also secured substantial funding at unicorn-level valuations, adding to the competitive landscape.

Alibaba’s latest claim highlights the broader trend of Chinese firms challenging Western counterparts in the AI space. Notably, according to an AFP report, Qwen2.5-Max is now available to developers through Alibaba Cloud services and can be accessed via Qwen Chat. Further, the system offers compatibility with OpenAI's API format, potentially simplifying adoption for organisations already using similar AI services.

(With inputs from Agencies)

Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.

Business NewsAIQwen 2.5 Max: AI battle heats up as Alibaba claims its new model outperforms Meta and DeepSeek in benchmarks
MoreLess
First Published:29 Jan 2025, 06:51 PM IST