How Open Models Win Against OpenAI and Anthropic

Open models face a critical economic challenge in differentiating themselves from vertically integrated closed-model giants like OpenAI and Anthropic. While open-weight models are gaining significant traction in research and enterprise usage—evidenced by the growth of inference startups like Fireworks and TogetherAI—they currently trail the frontier in cost-efficiency and performance. To remain competitive against the massive scale of closed-model labs, the open ecosystem must transition from merely chasing performance to providing highly adaptable, private, and domain-specific intelligence. Success depends on driving inference costs to marginal levels and enabling enterprises to retain ownership of their data and model weights. By focusing on specialized, on-premise deployments that closed labs cannot support, open models can secure a sustainable economic future and prevent a widening capability gap in the rapidly evolving AI landscape.