OpenAI Enters Token Price War! Luna Model Slashed by 80% as AI Giants Battle on Cost-Performance

The competition among artificial intelligence (AI) models has entered a new phase of price wars. OpenAI recently announced significant price reductions for its two latest AI models, GPT-5.6 Terra and GPT-5.6 Luna, with Luna’s prices slashed by up to 80% and Terra’s by 20%. Notably, this price adjustment comes just about three weeks after the models’ official launch, signaling that AI companies are accelerating their efforts to capture enterprise users through cost advantages.

OpenAI launched the next-generation GPT-5.6 series on July 9, including the flagship Sol, the work-focused Terra, and the high-speed, high-cost-performance Luna. Sol is designed for advanced applications such as programming, biology, and cybersecurity. Terra targets enterprise daily operations, while Luna is positioned as a low-cost option for high-volume usage.

According to OpenAI’s new pricing, the GPT-5.6 Luna API now costs $0.20 per million input tokens and $1.20 per million output tokens. The GPT-5.6 Terra is priced at $2.00 per million input tokens and $12.00 per million output tokens.

OpenAI stated that this adjustment will reduce the number of credits consumed when using Terra and Luna models, while the pricing and quota budgets for the Sol model, ChatGPT, and Codex subscriptions remain unchanged.

Corporate Cost Control Pressures AI Vendors

OpenAI’s price cut reflects heightened sensitivity to AI deployment costs in the enterprise market. As companies deploy generative AI at scale, computing expenses are rising rapidly. Without clear returns on investment, enterprises may hesitate to adopt high-priced models.

Industry observers note that OpenAI not only needs to meet enterprise demand for low-cost AI tools but also faces competitive pressure from tech giants like Google and Microsoft. These rivals have recently launched AI models with stronger price advantages, aiming to expand their market share in the enterprise sector.

OpenAI emphasized that its core strategy remains simultaneously enhancing model capabilities and operational efficiency, aiming for each new generation of intelligent models to accomplish more tasks at lower costs.

Token Price War Heats Up

The rapid decline in AI model prices indicates that the industry’s competitive focus is shifting from pure model capability to a comprehensive evaluation of performance and cost.

After ChatGPT’s debut in 2022, enterprises rushed to adopt AI at scale, even encouraging a phenomenon known as 'tokenmaxxing'—urging employees to use AI tools extensively without much cost consideration.

However, as AI usage increased, corporate AI spending surged. Some companies began reevaluating cost-effectiveness and implementing stricter cost controls.

Meanwhile, the rise of open-source AI models is reshaping the competitive landscape. Open-source models allow enterprises to download, modify, and run them on their own infrastructure, offering a high-cost-performance alternative.

Recently, Chinese AI startup Moonshot AI launched the open-source model Kimi K3, which performs competitively with leading U.S. models in certain tasks, prompting rapid responses from Silicon Valley AI firms.

Google and Anthropic Join the Competition

As the race for performance and cost efficiency intensifies, OpenAI’s competitors are also launching more cost-effective products. After Kimi K3’s release, Anthropic introduced a new model, Claude Opus 5, emphasizing its high performance and price competitiveness across most use cases. Although the model performs close to higher-tier products in programming and knowledge work tasks, its price is only half that of the premium models.

Microsoft (MSFT-US) CEO Satya Nadella recently emphasized the importance of high-cost-performance AI models during a quarterly earnings call. The company earlier launched a cybersecurity model marketed as 'low-cost, high-efficiency,' further highlighting the market’s focus on pricing efficiency.

Additionally, Google (GOOGL-US) launched several new models this month emphasizing cost advantages and claimed its Gemini 3.6 Flash has a lower per-task cost than Kimi K3. As major AI developers successively lower prices, the AI market is transitioning from a 'model capability race' to a full-fledged 'token cost competition.'

Analysts believe that future AI industry competition will no longer be solely about model parameter scale and reasoning ability. Instead, the key to capturing market share will be how effectively companies can reduce usage costs while maintaining performance.