Models

Nvidia open-sources a 30B mixture-of-experts model that activates only 3B parameters yet matches DeepSeek's 671B model on olympiad-level math and coding benchmarks.

The French AI startup ships its first text-to-speech model under open weights, offering enterprises voice cloning from three seconds of audio and 70-millisecond latency across nine languages.

OpenAI discontinues its Sora video generation app, ends a three-year Disney licensing deal, and redirects compute to a new model codenamed Spud ahead of a potential IPO.

An anonymous trillion-parameter AI model that topped OpenRouter usage charts and triggered widespread speculation about a stealth DeepSeek V4 launch turned out to be Xiaomi's MiMo-V2-Pro, signaling the smartphone and EV maker's escalating ambitions in the AI model race.

Meta postpones its next-generation Avocado AI model by at least two months after benchmarks show it trailing rivals, raising questions about the company's $135 billion AI investment.

Alibaba releases four open-source Qwen 3.5 Small models ranging from 0.8 to 9 billion parameters, delivering multimodal reasoning on consumer hardware while the 9B variant outperforms models up to 13 times its size on key benchmarks.

OpenAI releases its most capable model yet with autonomous computer operation and enterprise integrations, but the launch is overshadowed by Pentagon deal fallout and a reported 1.5 million user exodus.

Turing Award winner Yann LeCun's Paris-based startup raises the largest seed round in European history to develop world models, an alternative AI architecture that learns from physical reality rather than text prediction.

Alibaba releases Qwen3.5, a 397-billion-parameter open-weight model that activates only 17B parameters per pass, beats rivals on instruction-following benchmarks, and ships at 1/18th the cost of Gemini 3 Pro.

OpenAI's Pentagon deal sparked a consumer backlash that drove Anthropic's Claude to the top of Apple's App Store, before a surge in demand caused a worldwide outage.

DeepSeek's first major AI release in over a year arrives amid Anthropic allegations that the Chinese lab used 24,000 fake accounts and 16 million queries to extract Claude's capabilities.

Google DeepMind's Gemini 3.1 Pro launches with a 2.5x reasoning improvement over its predecessor, topping the ARC-AGI-2 benchmark at 77.1% and ranking first across most tracked evaluations.