OpenAI and Anthropic in price war as Chinese AI rivals gain ground
OpenAI and Anthropic are aggressively cutting prices on mid-tier AI models to compete with cheaper Chinese alternatives like Moonshot and DeepSeek OpenAI slashed GPT-5.6 Luna prices by 80%, while Anthropic launched Claude Opus 5 at half the price of its flagship Fable 5 Chinese "open" models have narrowed the performance gap, prompting US labs to shift from performance-only competition to price competition Enterprise customers are increasingly adopting usage-based billing and imposing AI usage c
Analysis
TL;DR
- OpenAI and Anthropic are aggressively cutting prices on mid-tier AI models to compete with cheaper Chinese alternatives like Moonshot and DeepSeek
- OpenAI slashed GPT-5.6 Luna prices by 80%, while Anthropic launched Claude Opus 5 at half the price of its flagship Fable 5
- Chinese "open" models have narrowed the performance gap, prompting US labs to shift from performance-only competition to price competition
- Enterprise customers are increasingly adopting usage-based billing and imposing AI usage caps to control rising costs
- The price war coincides with OpenAI and Anthropic planning trillion-dollar IPOs, adding investor pressure to demonstrate ROI on AI spending
Why It Matters
This price war signals a fundamental shift in the AI industry's competitive dynamics, as US labs can no longer rely solely on performance advantages to retain customers. For AI practitioners and enterprises, it highlights the growing importance of cost-efficiency in model selection and the real threat posed by Chinese open-source alternatives. The trend also reflects broader industry pressures around ROI expectations as companies prepare for major public offerings.
Technical Details
- OpenAI reduced GPT-5.6 Luna pricing from $1 to $0.20 per million input tokens and from $6 to $1.20 per million output tokens, representing an 80% reduction
- Anthropic launched Claude Opus 5 at $5 per million input tokens and $25 per million output tokens, positioned at half the price of Fable 5
- Artificial Analysis benchmarks showed Anthropic's Opus 5 at "medium" effort delivered similar performance and cost per task to Moonshot's Kimi K3 at "max" effort
- OpenAI's GPT-5.6 Luna at "max" effort performed comparably to DeepSeek's V4 Flash at "max," but cost nearly twice as much per task
- Token pricing remains an imperfect comparison metric, as more capable models may complete tasks with fewer tokens or attempts, and "effort" settings further complicate cost calculations
Industry Insight
- US AI labs are adopting a "cut the middle, defend the top" pricing strategy, suggesting they view mid-tier models as most vulnerable to Chinese competition while maintaining premium pricing for frontier capabilities
- Enterprises should evaluate models based on total cost per task rather than headline token prices, accounting for effort settings and task completion efficiency
- The narrowing performance gap between US and Chinese models indicates Chinese labs are approaching technological parity, making price and ecosystem lock-in the primary differentiators going forward
Disclaimer: The above content is generated by AI and is for reference only.