Why the Rise of Open-Source AI Isn’t Hurting Anthropic Yet
Despite surging usage of open-source models like DeepSeek, spending on frontier models, especially Anthropic’s, remains high because the market is expanding rapidly and these models still dominate new use-case discovery.
Decagon CEO Jesse Zhang has proposed a theory that open-source and frontier AI models are not direct competitors but rather operate in different phases of the same lifecycle. In his view, expensive frontier models prove out use cases, which are then handed off to cheaper open-source alternatives as they mature.
Data from Vercel’s AI gateway dashboard shows that in the past week, DeepSeek has taken the lead in token volume, processing just over a third of all tokens passing through the company’s infrastructure. Z.ai’s GLM-5.2 model moved into fourth place. However, when looking at total AI spend, Anthropic still accounts for more than half of all spending on the platform, though its share has dipped slightly in the past month due to Anthropic’s own price increases.
OpenRouter tells a similar story: DeepSeek V4 Flash processes 5.3 trillion tokens weekly, while the most popular frontier model, Opus 4.8, handles just over 2 trillion. Yet Opus 4.8’s average token cost is roughly 23 times higher ($1.37 per million tokens vs. $0.06), so it likely captures the lion’s share of spending. These figures don’t yet include Nvidia’s new Nemotron model, which is poised to leap ahead thanks to Nvidia’s strong connections and the model’s adaptability.
Zhang argues that frontier labs will keep owning discovery, while open source will increasingly own production. Another explanation is that many use cases are so challenging that they cannot be fully replaced by cheaper alternatives. This two-tiered model economy may become a stable feature of the AI landscape, with frontier models retaining premium pricing while open-source models handle higher volumes.
