OpenAI Cuts GPT‑5.6 Sol Pricing by More Than 20%—and the Real Competition Is Cost per Useful Result

OpenAI has reduced the API and credit pricing of GPT‑5.6 Sol by more than 20% for three months, placing a fresh spotlight on an increasingly important measure in artificial intelligence: the cost of producing a useful result.

Frontier models have usually been compared through benchmarks, speed and maximum capability. Businesses, however, experience AI through budgets. A model may be extremely powerful, but the practical question is how much reliable work it completes for each pound spent.

Lower prices can expand experimentation. Small companies may be able to analyse larger document collections, build richer customer tools or run more complex agent workflows without treating every request as a premium event.

Price reductions also intensify competition across the market. Model providers must now balance intelligence, latency, reliability, security and cost. A cheaper system is not automatically better if it requires repeated corrections; an expensive system may still be economical when it succeeds on the first attempt.

The sensible business response is measurement. Organisations should track the complete cost of a workflow, including human review, failed attempts, tool calls and the value of time saved.

AI access grows when capability becomes affordable. The important question is not which model has the most dramatic demonstration, but which one turns intelligence into dependable value at a sustainable price.