OpenAI is reducing the cost of using its most powerful model, GPT-5.6 Sol, thereby intensifying competition for companies that are increasingly mindful of the costs of implementing generative AI.
The new API prices are $4 per million input tokens and $20 per million output tokens. Previously, these were $5 and $30 respectively, representing a 20 per cent reduction for input data and a one-third reduction for model responses. The promotional rates are set to remain in place for around three months. The subscription prices for ChatGPT Plus, Pro and Business remain unchanged.
This is yet another price cut by OpenAI in a short space of time. At the end of July, the company reduced the prices of the GPT-5.6 Terra and Luna models, cutting the cost per token for the latter from $1 to $0.20 per million tokens.
These changes show that competition between model providers is increasingly shifting from benchmarks to the economics of deployment. Anthropic’s Claude Opus 5 currently costs $5 per million input tokens and $25 per million output tokens, whilst the more advanced Claude Fable 5 costs $10 and $50 respectively.
For businesses, a lower price primarily means the ability to run a greater number of queries and AI agents within the same budget. This is particularly significant in coding, document analysis and process automation, where token consumption rises rapidly as the scale of application increases.
At the same time, the three-month promotional pricing period suggests that OpenAI is testing not only demand but also customers’ sensitivity to the cost of the models. If competitors respond with further price cuts, the cost of inference could become one of the main battlegrounds in the race for the enterprise AI market.
