OpenAI on Friday announced it would cut the price of its advanced AI model GPT-5.6 Sol, offered to developers, by more than 20% for three months. The move comes as OpenAI faces increasing competition from AI models by Anthropic and Chinese rivals.
According to OpenAI, the discounts apply to use of GPT-5.6 Sol via the API and will be applied to eligible credit plans in the ChatGPT Work AI proxy product as well as the Codex programming tool. However, subscription prices for Pro, Plus, and Business remain unchanged.
Under the new pricing, standard use of GPT-5.6 Sol for short text will cost $4 per one million input tokens and $20 per one million output tokens. Previously, input and output tokens cost $5 and $30 per one million tokens, respectively.
Thus, the cost of input tokens has decreased by 20% and the cost of output tokens is about one-third lower. This price reduction could lower operating costs for applications that rely heavily on content generation and model-generated responses.
OpenAI also cut prices for its smaller models
The price cut for GPT-5.6 Sol follows OpenAI’s reductions for its smaller models late last month. As part of making its models more competitive, the company cut the price of the mid-tier GPT-5.6 Terra by 20% and the cheaper Luna by 80%.
These measures come as competition in the advanced AI model market intensifies. In this market, the cost of using models has become a key factor for developers and companies when choosing the technology to build applications and services.
Competition with Anthropic’s models
Anthropic’s model pricing also reflects the competitive pressure OpenAI faces.
Anthropic offers its advanced Claude Fable 5 at $10 per one million input tokens and $50 per one million output tokens. The price for Claude Opus 5 is about $5 per one million input tokens and $25 per one million output tokens.
Based on a comparison of prices provided in Reuters data, after the price cuts GPT-5.6 Sol will be cheaper in terms of input and output token costs than both of those models.