Summary of Key Points
Recently, the AI industry has experienced a stark contrast in strategies: OpenAI has upgraded its model (GPT-5.6 Luna) for free users, removed restrictions on the number of pure-text conversations, and added a “deep thinking” feature; meanwhile, DeepSeek has announced a significant increase in the prices of its API services. These seemingly opposite moves actually point to the same reality within the industry—the cost structure of AI is being restructured. The industry is transitioning from a phase where companies were spending heavily to expand their scale to one where commercialization becomes possible with more controlled costs, signaling a reshuffle of the landscape. OpenAI is focusing on acquiring users and data at low costs, while DeepSeek is forced to raise prices due to shortages in computing power. Both companies are redefining the boundaries between free and paid services, with the ultimate test being their ability to create real value.
1. OpenAI’s Free Service: Not Charity, but a Smart Business Move
OpenAI’s recent decision to make its services more accessible for free is not out of charity; it serves three strategic purposes:
- Reduced Cost of Pure Text Processing: The price of the GPT-5.6 Luna API has been reduced by 80% in just three weeks (0.2 USD per million characters input), making the cost of free conversations almost negligible. By offering a free tier, OpenAI can gain more traffic and build a positive reputation at a much lower cost.
- Gathering Data from 1 Billion Users: With 1 billion people using ChatGPT each week, each additional free conversation provides valuable feedback for model training and interaction optimization. This data is far more valuable than the cost of charging for the service.
- Strengthening Market Positioning: Apple will launch an AI version of Siri next month, and domestic platforms like DouBao and Kimi are also offering free services. If OpenAI continues to insist on a paid model only, it may lose users to these competitors. By making its free tier more attractive, OpenAI aims to establish a user base that is difficult for others to compete with.
In short, OpenAI provides a free chat platform but charges for advanced services like image generation and file processing, using the free tier as a means to attract users before generating revenue from higher-value offerings.
2. DeepSeek’s Price Hike: Panicked by Small Teams and B2B Service Providers
DeepSeek recently cut prices by 75% three months ago but has now raised them significantly due to a shortage of computing power. Its models used to process 8 trillion characters per day, and the high traffic caused capacity issues. Expanding the infrastructure requires purchasing more GPUs, which are in short supply and expensive, forcing the price increase.
The main affected groups include:
- Companies That Can Pass On Costs: Financial risk management and pharmaceutical research, which purchase AI services for tangible benefits (reducing bad debts, shortening development cycles) and are less sensitive to price increases.
- Companies That Cannot Pass On Costs: Consumer-facing AI writing and companionship services, where the monthly fee is low, and price hikes directly affect profits. These companies may either bear the increase, switch to cheaper models, or cease operations.
- B2B Service Providers with Fixed Revenue: Those with annual contracts see their costs rise, leading to greater losses as they try to maintain their business.
DeepSeek’s price hike serves as a warning to those using low-cost AI models, indicating that the cost of processing text (in tokens) is not indefinitely affordable.
3. The Reconfiguration of AI Industry Costs
The contrasting strategies of OpenAI and DeepSeek reflect a broader shift in the industry’s approach to calculating costs:
- Redefining Free and Paid Services: OpenAI offers free chat as an entry point and charges for more advanced features; DeepSeek, after a period of free trials, now requires payment for its productivity tools.
This transition marks a move from an era of aggressive cost-cutting to one where companies must focus on providing valuable services. The implications are significant:
- Model Manufacturers: The strategy of offering “infinite low prices” is no longer effective. With limited computing power and difficult financing, companies need to offer valuable solutions to stay competitive.
- Application Developers: Choosing an AI model is no longer just a technical decision but also involves financial considerations. Only those who can effectively pass on the costs to their customers will survive.
- End Users: Free chat services will become more widespread (like search engines), while paid services for complex tasks (PPT creation, coding) will become more expensive, as the value created by AI increases.
4. Who Will Be Eliminated?
The reevaluation of AI costs puts three types of players at risk:
- Shell Companies: Those that rely on low-cost APIs without their own data, use cases, or user bases will lose profits when API prices rise.
- Products with Inefficient Cost Management: Companies that promise unlimited AI features without optimizing costs (e.g., by not using caching) will suffer significant losses as their scale expands.
- Developers Dependent on Single Models: Those who are tied to a single API will face higher costs and have no alternative options.
Only those who understand the balance between cost and value—knowing what to offer for free (as an entry point) and what to charge for (value-added services)—and can effectively convert token usage into customer value, while being flexible in model selection and optimizing costs, will emerge victorious in this new landscape.
In Conclusion
The “free lunch” in the AI industry is becoming a paid service. Some functions (like basic chat) will remain free, but others (like advanced tasks) will become more expensive. The criterion for distinguishing between free and paid services is simple: how much real value they can create for users. This reshuffle will ultimately favor those companies that can balance costs and deliver meaningful value to their customers.