Open-weight/open-source AI models are gaining traction in production use, led by Chinese firms, as cheaper, customizable options threaten frontier models' dominance and raise questions about where value sits in enterprise AI.
Open-source AI isn’t yet hurting Anthropic; frontier labs still capture the premium spend and Anthropic remains the dominant share of platform spend, implying a two-tier AI economy with resilience for incumbents.
Vercel's Guillermo Rauch argues for decoupling models from agents and highlights scale, governance, and open-protocol competition, citing 6 million deployments per day and over 1 trillion tokens through its AI gateway coupled with data-safety tools like Eve and Sandbox.
GPT-5.5 allegedly hallucinates 3x more than MIT-licensed GLM-5.2, per a Hacker News discussion.
Subscribe for real-time topic updates and unlimited access to our intelligence platform.