Urgent.News

One page, thousands of outlets. See who else covered it.

Editions

AI

Meta and Google just got beaten by an AI giant you might not expect

Alibaba’s Qwen has become a massive force in open AI, crossing 3 billion downloads in just six months and leaving some of Silicon Valley’s biggest names far behind.

Meta and Google just got beaten by an AI giant you might not expect

The AI race is usually dominated by well-known names such as OpenAI, Google, Meta, and Anthropic. However, Alibaba's Qwen family of open-weight AI models has emerged as a formidable competitor, garnering 3 billion global downloads within the last six months. This achievement surpasses the download counts of major players like Meta and Google's open models, which have recorded approximately 418 million and 227 million downloads, respectively.

Alibaba's Qwen is not just a single popular model; the company has released over 460 Qwen models, leading to the creation of more than 300,000 derivative models by developers. The open nature of these models allows for extensive customization, fine-tuning, and integration into new AI products, making each download potentially transformative.

This widespread adoption is contributing to a growing open AI ecosystem, with Qwen becoming a preferred choice for developers seeking to fine-tune or deploy AI solutions. Alibaba's cloud business provides an additional advantage by expanding Qwen's availability to enterprise customers in Southeast Asia and Africa, potentially outpacing competitors from the US.

The AI rivalry between China and the US is intensifying, with other Chinese firms like DeepSeek, Moonshot AI, and MiniMax also developing advanced models. Meta and Nvidia have recently introduced new open AI models, further expanding the options available to developers. For Alibaba, achieving 3 billion downloads highlights Qwen's substantial global reach.

The next major AI competition may revolve around whose model developers choose to build upon, rather than which model is the most sophisticated.

Written by urgent.news from Digital Trends's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at digitaltrends.com →

More in AI

Unpopular Opinion: Why I’m an AI Skeptic

With all the hype in the past several years around AI (or more specifically GenAI), I'm not afraid to say – I'm an AI skeptic.

  • AI hype is driven by irrational enthusiasm, not solid evidence
  • Current AI lacks true self-learning and emotional capabilities
  • Enterprise benefits from AI are uncertain until technology matures

How do you catch it when a model update changes your agent's tool calls?

Your agent calls get_weather(city="London") . The provider ships a new model version. Now it calls get_weather(location="London, UK") , your downstream parser breaks, and nothing in CI told you.

  • Toolcontract library detects model update changes to tool calls
  • Golden set of expected tool calls serves as contract
  • Library supports strict, unordered, subset, superset trajectory matching

Semantic Cache in AI Tokenomics

Introduction A few weeks ago I was looking at logs for a support bot that answers the same handful of questions all day long.

  • Semantic cache addresses redundant AI model calls for semantically similar queries.
  • Semantic cache uses embeddings, similarity scores, and thresholds for meaning-based caching.

More from Sunday 16 August →