Alibaba upgrades flagship Qwen3 model to outperform OpenAI, DeepSeek in maths, coding
The new Qwen3-235B-A22B-Instruct-2507-FP8 is an open-source model that achieved 'significant improvements in general capabilities, including instruction following, logical reasoning, text comprehension, mathematics, science, coding and tool usage', according to a Tuesday update on artificial intelligence (AI) community HuggingFace and ModelScope, Alibaba's open-source platform. Alibaba owns the Post.
It outperformed some rivals in certain assessments, such as the 2025 American Invitational Mathematics Examination, where the new Alibaba model scored 70.3. By comparison, DeepSeek-V3-0324, the most recent version of the foundational model that was released in March, scored 46.6 while OpenAI's GPT-4o-0327 scored 26.7.
As for coding capabilities, the new Qwen secured 87.9 points from the MultiPL-E benchmark, slightly higher than 82.2 and 82.7 from the DeepSeek and OpenAI models above, respectively, though it lagged behind Claude Opus 4 Non-thinking, from Anthropic, which scored 88.5.
Alibaba's new release was an upgrade from the Qwen3-235B-A22B-FP8. But it only supports non-thinking mode, where an AI system provides a direct output without the explicit reasoning steps or chain of thought that a thinking model might employ. As a result, its content length was boosted eightfold to 256,000 tokens, making it able to handle longer texts in a single conversation.
Also on Tuesday, Alibaba said a Qwen model with 3 billion parameters would be integrated into HP's smart assistant 'Xiaowei Hui' on its personal computers in China, enhancing capabilities including drafting documents and summarising meetings.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
South China Morning Post
40 minutes ago
- South China Morning Post
China's quant hedge funds stock up on talent – at US expense – to fuel expansion, AI use
Yuan Yu, founder of Shanghai-based quantitative hedge fund Mingshi, recognised an opportunity when several intern prospects studying in the US said they were struggling to complete their PhDs amid university funding cuts and stricter visa policies under President Donald Trump. 'They told us that their supervisor's funding had been cut, so they couldn't continue their studies,' said Yuan, whose investment firm manages US$2.5 billion in assets. 'They felt lost and didn't know what to do. This prompted us to consider extending our internship offers to full-time job offers.' By seizing the chance to snap up the candidates, Mingshi echoed a larger trend in China's rapidly evolving quant hedge fund sector. The US upheaval in academia – and the government's antipathy towards foreign students – is helping such funds secure the brain power that they will need to harness China's breakthroughs in artificial intelligence (AI) and fuel their expansion beyond the domestic market. Companies like Mingshi often compete against local rivals and larger, more established global giants for the world's top scientists and engineers, many of whom would have stayed in the US in the past. 'We are hungry for top talent,' Yuan said. 'In the past two to three years, we have been offering salaries that are higher than some of the top US firms.' The sector has been growing rapidly. The yuan-denominated market is now the world's second largest, with a combined capitalisation of more than US$11 trillion, feeding demand for financial products aimed at wealth preservation or hedging. Quant products managed by domestic hedge fund firms totalled 837 billion yuan (US$117 billion) at the end of 2024, with 768 billion yuan invested in stocks, according to Citic Securities. Such products from mutual-fund firms, mostly index-reinforcing offerings, held 295 billion yuan, it said.
South China Morning Post
2 hours ago
- South China Morning Post
China airport lounge criticised for demanding proof of assets worth US$2.8 million for entry
A VIP lounge at a Chinese airport has ignited considerable controversy after mandating proof of assets worth at least 20 million yuan (US$2.8 million) for entry, citing a desire to 'ensure the quality of service.' Advertisement On July 17, a Chinese netizen shared online that, as a client of China Merchants Bank's 'Golden Sunflower' programme, she inquired about accessing the international First Class Lounge at Chengdu Tianfu International Airport in Sichuan province in southwestern China. She was informed that, in addition to using reward points, she would also need to provide proof of assets amounting to at least 20 million yuan. The netizen explained that she had bought an economy class ticket but aspired to rest in the international First and Business Class Lounge at Terminal 1. She wrote: 'I arrived at the airport early and searched online for available lounges. I noticed that Golden Sunflower clients could use them, and since I happen to be one, I called for clarification. But then I learned I needed to verify 20 million yuan in assets! Moreover, property and vehicles don't count! Poverty really limits my imagination!' Many banks commonly include airport lounge access as a benefit in their credit card and banking product redemption programmes to attract and retain clients. Photo: Shutterstock She continued: 'I'm baffled as to how they determined this 20 million yuan and 9 points threshold. Anyone with such wealth would likely just purchase a business-class ticket outright. Why would they bother redeeming points for lounge access?' According to publicly available information, Golden Sunflower cards typically require an average monthly balance of at least 500,000 yuan (US$70,000) to open, with airport lounge access being a primary benefit promised to cardholders. In response to the requirements, airport staff explained: 'If you've bought a first or business class international ticket, you can access the lounge directly. If not, you can pay 600 yuan (US$85) for individual use. For China Merchants Bank cardholders, verifying assets over 20 million yuan is necessary. Lounge access is by reservation only, and cardholders must contact the issuing bank to verify eligibility and complete the booking process.' Passengers flying in first or business class typically enjoy VIP lounge access, enhancing their overall travel experience. Photo: Shutterstock China Merchants Bank later confirmed that this rule has been effective since June 2024.
South China Morning Post
13 hours ago
- South China Morning Post
China promised the Philippines billions in development aid. Why did it fall so short?
China pledged US$30.5 billion in development aid to the Philippines between 2015 and 2023 – the most for any Southeast Asian country – but only a sliver of that funding ever arrived, according to new data from an Australian think tank report. Advertisement Of the total pledged, just US$700 million was actually disbursed – a shortfall analysts attribute to derailed infrastructure projects, changing political winds in Manila and rising tensions with Beijing. These factors have not only stalled flagship ventures under the Belt and Road Initiative but also cast doubt on the long-term viability of Chinese development finance in the region. The report by the Sydney-based Lowy Institute, released on Sunday, found that while the Philippines received the highest total commitment from China among Southeast Asian nations, it ranked near the bottom in actual disbursements. Indonesia, by contrast, received and spent US$20.3 billion out of the US$20.7 billion Beijing had pledged, mostly on energy and transport projects. Philippine President Rodrigo Duterte (left) and his Chinese counterpart Xi Jinping in Beijing in 2017. Duterte pursued closer ties with Beijing through a wave of high-profile infrastructure agreements. Photo: AP The bulk of China's pledged financing to the Philippines was made during the administration of former president Rodrigo Duterte , who held office from 2016 to 2022 and pursued closer ties with Beijing through a wave of high-profile infrastructure agreements.
