Why the rise of open source AI isn’t hurting Anthropic … yet
Summary
Jesse Zhang's theory posits that open-source and frontier AI models are not competitors but sequential phases in an AI deployment lifecycle. Expensive frontier models initially validate use cases, which then migrate to cheaper open-source alternatives as they mature. Despite this transition, overall spending on frontier models remains high due to the rapid emergence of new AI-addressable tasks. Data from Vercel's AI gateway dashboard shows DeepSeek leading in token volumes, processing over a third, while Anthropic still commands over half of the total AI spend. Similarly, OpenRouter data indicates DeepSeek V4 Flash processes 5.3 trillion tokens weekly compared to Opus 4.8's 2 trillion, but Opus 4.8's cost of \$1.37 per million tokens is 23 times higher than V4 Flash's 6 cents, suggesting frontier models retain the majority of spending. This indicates a stable, two-tiered AI economy where frontier labs drive discovery and open source handles production, maintaining premium token prices for top providers.
Key takeaway
For AI Product Managers evaluating model deployment strategies, recognize that frontier models like Anthropic's Opus 4.8 excel in initial use case discovery despite their higher cost. You should plan to utilize these for validating new applications, then transition mature, high-volume workloads to cost-effective open-source alternatives such as DeepSeek V4 Flash. This two-tiered approach optimizes both innovation and production efficiency, ensuring your budget supports both exploration and scalable operations without sacrificing performance.
Key insights
Frontier AI models drive discovery, while open-source models increasingly own production in a two-tiered AI economy.
Principles
- AI models follow a lifecycle: discovery then production.
- Frontier models maintain premium pricing for early-stage tasks.
- Market growth sustains frontier model spend.
In practice
- Evaluate frontier models for initial use case validation.
- Transition mature AI workloads to open-source alternatives.
- Monitor token volume vs. spend for cost optimization.
Topics
- Open-Source AI
- Frontier Models
- AI Economy
- Model Deployment
- Token Pricing
- Anthropic
- DeepSeek
Best for: AI Engineer, Machine Learning Engineer, NLP Engineer, Director of AI/ML, AI Product Manager, Investor
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.