The AI Price War Is Here, And Chinese Models Are Forcing America’s Hand
Claude Opus 5.5 and GPT-6 Sol and Luna arrived within hours of each other this week, both cheaper than before, as cost-conscious businesses turn to less expensive rivals from China.
OpenAI and Anthropic spent two years telling everyone their AI models were getting smarter. This week, they spent it telling everyone their models got cheaper.
On the same day, Anthropic released Claude Opus 5.5, priced about 40 percent lower than its predecessor while matching the performance of its flagship Fable 5.1 model on most tasks. Hours later, OpenAI launched GPT-6 Sol and GPT-6 Luna, cutting prices by half compared to its previous GPT-5.6 lineup.
Why the sudden discount
For most of the AI race, these companies competed on intelligence. Now they’re competing on cost, and the pressure is coming from China. DeepSeek’s newest model, V4.1 Flash, jumped to number one on OpenRouter’s rankings this month, with usage up 172 percent in a single week. Moonshot AI and other Chinese labs are offering similarly capable, open-weight models for a fraction of the price.
Businesses have noticed. Many companies, including some in China, are now calculating “cost per task” instead of paying extra for raw power they may not fully need.
Cutting prices without losing the top
Interestingly, OpenAI and Anthropic aren’t discounting everything equally. Their most advanced, top-tier models are staying expensive, while cheaper mid-tier versions absorb the price cuts. One industry analyst summed it up simply: the U.S. labs are cutting the middle to defend the top.
There’s also a financial bet behind this. Falling prices tend to push usage up, and several analysts believe total AI spending could keep rising even as per-token costs fall.
The real question is whether that bet pays off before investors, especially with Anthropic reportedly eyeing an IPO, start asking if cheaper AI still means profitable AI.
Get the latest tech news on WhatsApp
Comments are closed.