Meta's new open-weight model is pitched as a way to run AI agents on a laptop, but available reporting leaves its hardware and performance limits unclear.

Meta is reportedly releasing an open version of its most powerful AI model, challenging OpenAI and Anthropic while reshaping access for builders and enterprises.

Meta has reportedly launched Muse Glimmer, a local AI model, while Mark Zuckerberg renews Meta’s open-weight strategy amid competition from China.

Alibaba has introduced its largest AI model while DeepSeek highlights ultra-low costs, intensifying pressure on providers to deliver more for less.

Alibaba has launched Qwen3.8-Max, described as its largest AI model, putting fresh pressure on rivals in China’s fast-moving model market.

LG AI Research has unveiled a 750-billion-parameter model, putting scale at the center of South Korea’s competition to build advanced AI systems.

Liquid AI released LFM2.5-Encoder models on Hugging Face, pitching faster long-context CPU inference for document-scale NLP without larger hardware.

SK Telecom introduced a second A.X K2 Korean-language model, signaling a deeper push into domestic enterprise AI with stronger reasoning claims.

Moonshot AI has released Kimi K3 weights and infrastructure, widening access to a near-frontier open model as questions persist over its real capabilities.

Moonshot AI’s latest Kimi model has revived debate over open Chinese AI, as benchmark claims, market jitters, and policy fears collide.

Moonshot AI has unveiled Kimi K3, a massive open-weight model that media reports say reaches frontier-level results and raises the stakes for builders and rivals.

Thinking Machines Lab’s first model, Inkling, is now available on Databricks, signaling an early push to reach enterprise AI teams through customization.

Thinking Machines has launched an open-weight AI model, signaling Mira Murati’s startup plans to compete where builders want deployable models.

Reports say Mistral has launched a robotics model as its valuation approaches $23 billion, signaling a broader push beyond core foundation models.

Meta introduced Muse Spark 1.1 and a public preview of the Meta Model API, pairing a 1 million-token context window with pricing that undercuts U.S. rivals.

Nous Research has released NousCoder-14B, an Apache 2.0 coding model with an open training stack, as demand for AI programming tools intensifies.

SpaceXAI has launched Grok 4.5, positioning the model as a cheaper, faster rival for coding and knowledge work ahead of a new OpenAI release.

Mistral has launched Robostral Navigate, an 8B robotics model that steers robots with one RGB camera, signaling a broader push into embodied AI.

Mistral AI has introduced Leanstral 1.5, an Apache-2.0 Lean 4 code agent model that reportedly solves 587 of 672 PutnamBench problems.

Meta debuts Muse Spark, its first major LLM from Meta Superintelligence Labs led by chief AI officer Alexandr Wang, targeting rivals Google and OpenAI.