OpenAI launched GPT-5.5 with doubled pricing compared to GPT-5.4, raising input tokens from $2.50/M to $5.00/M and output tokens from $15/M to $30/M. However, an analysis by OpenRouter reveals that actual cost increases for users range from 49% to 92% depending on prompt length, as the model's improved token efficiency partially offsets the nominal price hike.
Short Prompts See 92% Cost Increase, Long Prompts 49-85%
OpenRouter's detailed cost analysis shows the practical impact varies significantly by use case. Users with short prompts under 2K tokens experience the highest actual cost increases at 92%. Medium prompts between 2K-10K tokens see 69% increases, while long prompts over 10K tokens experience 49-85% increases. The variation stems from GPT-5.5's reduced verbosity, producing 19-34% fewer completion tokens for longer prompts compared to GPT-5.4.
Token Efficiency Gains Don't Fully Offset Doubled Per-Token Costs
While GPT-5.5 demonstrates improved efficiency by generating fewer tokens for equivalent outputs, this gain doesn't fully compensate for the doubled per-token pricing. OpenRouter conducted the analysis by comparing users who switched from GPT-5.4 to GPT-5.5, controlling for the same tokenizer family to ensure accurate comparisons. The analysis provides transparency on real-world cost implications that many users may not have anticipated from the nominal price change alone.
GPT-5.5 Pro Targets Complex Professional Workloads
GPT-5.5 is positioned as OpenAI's frontier model for complex professional workloads, featuring stronger reasoning, higher reliability, and improved token efficiency on difficult tasks. A Pro version is also available at significantly higher rates of $30/M input and $180/M output tokens, optimized for deep reasoning on complex, high-stakes workloads. OpenAI has not publicly disclosed specific reasons for the price doubling.
Key Takeaways
- OpenAI doubled GPT-5.5 pricing to $5.00/M input and $30/M output, up from $2.50/M and $15/M respectively
- Actual user cost increases range from 49% to 92% depending on prompt length, not the full 2x nominal increase
- Short prompts under 2K tokens see the highest cost impact at 92%, while long prompts see 49-85% increases
- GPT-5.5 produces 19-34% fewer completion tokens for longer prompts, partially offsetting the price increase
- A Pro version is available at $30/M input and $180/M output for complex, high-stakes reasoning tasks