Open Source vs. Closed Source "Money Grab Battle": Price Differences Reach 40x, Open Source Models Divert Leading Tech Giants' Billions in Revenue

According to Beating Monitoring, after the release of 25 open-source weight models including NVIDIA Nemotron 3 Ultra and Google Gemma 4 12B, tech investor Chamath Palihapitiya pointed out that the gap in capabilities between cutting-edge open-source and closed-source models is rapidly narrowing, but the cost gap for calls still remains huge.

Based on a monthly consumption of 1 billion input tokens and 1 billion output tokens per enterprise, the cost of GPT-5.5 Pro reaches up to $105k, Claude Opus 4.8 costs $30k, while DeepSeek R1 only costs $2,740, making GPT-5.5 Pro nearly 40 times more expensive. Chamath warned that most company CEOs are unaware that their technical teams, lacking governance and auditing, are directly defaulting to the most expensive large model APIs, leading to budget overruns.

As model routing control platforms like Software Factory, an AI-native software development lifecycle orchestration platform he co-founded and serves as CEO, become increasingly popular, enterprises will shift to model-agnostic architectures, defaulting to divert large-scale inference to DeepSeek, routing high-end proxies to Claude Opus, and only calling GPT-5.5 Pro on demand when clear high-value incremental results are produced. Chamath predicts that refined routing will significantly slow the API revenue growth of frontier labs like OpenAI and Anthropic, while revenue from open-source and low-cost inference ecosystems will experience explosive growth.