Introduction
The launch of GPT-5.5 by OpenAI has brought about a significant price increase that has caught the attention of many users. So, what is the real cost of this upgrade? We dive into an in-depth analysis of cost changes and model performance.
Cost Comparison
GPT-5.5 launched with a 2x price increase over GPT-5.4. Input tokens rose from $2.50 per million to $5.00, while output tokens increased from $15 to $30 per million. OpenRouter conducted a cost-benefit analysis to understand the impact of these changes.
Usage and Actual Cost
The analysis revealed cost increases between 49% and 92% depending on usage. Longer prompts benefit from shorter completions, reducing the impact of the price hike. However, shorter prompts experience a more pronounced cost increase without significant reductions in completion length.
Analysis Methodology
OpenRouter used a "switcher cohort" approach to compare usage before and after the GPT-5.5 launch. By analyzing request logs, they were able to normalize costs per million OpenRouter tokens, providing a precise overview of costs.
Impact on Different Prompt Sizes
Here are the results:
- < 2K tokens: Cost increase of 92%
- 2K – 10K tokens: Increase of 69%
- 10K – 25K tokens: Increase of 51%
- 25K – 50K tokens: Increase of 62%
- 50K – 128K tokens: Increase of 49%
- 128K+ tokens: Increase of 85%
Conclusion
Transitioning to GPT-5.5 involves a substantial cost increase, especially for users with shorter prompts. For those using longer prompts, the cost is partially offset by shorter completions. Ultimately, the impact heavily depends on each user's specific usage.
Let's discuss your project in 15 minutes.