Two American tech giants are pushing back against calls for restrictions on open-weight AI by releasing their own free-to-download models this week, aiming to compete with popular offerings from Chinese labs. Meta launched Muse Glimmer on Monday, with CEO Mark Zuckerberg also pledging to open the weights for its upcoming Muse Spark 1.2 model. The next day, Nvidia debuted Nemotron 3.5 Lightning, which it calls "truly open source" because it publishes training datasets, techniques, and model weights.
The moves follow a July 24 open letter from a consortium of U.S. tech companies urging policymakers to avoid "premature restrictions" on open-weight models, even those from China. The letter argued that open-weight AI can expand opportunity and strengthen competition, and defended the practice of distillation as "a widely used technique for model improvement, evaluation, and validation." Critics, however, have raised national security concerns about Chinese models and view distillation as potential intellectual property theft.
Box CEO Aaron Levie, a signatory of the letter, called Zuckerberg's plan for Muse Spark 1.2 a "very big deal," noting it rivals top foundation models from Anthropic and OpenAI. He said the new models are smaller and designed for laptops and on-device digital agents. "There's a very firm flag in the ground that America will have near-frontier open-source models," Levie said. He added that organizations wary of using non-domestic models—such as government agencies and large banks—would be more inclined to experiment with Meta's offering.
Meta has tried open-source before with Llama, but the April 2025 release of Llama 4 disappointed developers. The company then spent billions overhauling its AI unit under Scale AI CEO Alexandr Wang, who has since rolled out proprietary Muse-branded models for revenue. Some developers remain skeptical. Umesh Sachdev, CEO of business AI startup Uniphore, said Meta's shift away from open weights earlier burned bridges with third-party developers. "The emotion of my developers at Uniphore, they almost feel betrayed," he said, but he still supports domestic competition because it lowers token costs and drives innovation.
Forrester analyst Charlie Dai called Meta's latest move "strategically important" for restoring a major U.S. frontier AI vendor to the open ecosystem, but said the company must "prove it can cultivate a durable ecosystem beyond releasing competitive models."
Source: CNBC