NewsBite analysis · AI, Business, Cloud
The Decoder
OpenAI reportedly cut response costs for guest ChatGPT users by more than half
OpenAI has made substantial improvements in reducing the inference costs for guest users of ChatGPT by over half. This optimization, reported by individuals close to the discussions, involves a significant reduction in the number of Nvidia GPUs required to serve these users. Although guest users have limited access to features, this shift could allow for enhanced service capacity and efficiency. The effectiveness of these changes for fully registered users remains uncertain, highlighting an area for future exploration as OpenAI seeks to balance costs with performance improvements.
MonitorConfidence: MediumShort term horizon
Published 30 Jun 2026, 18:10 UTC1 min read
Visit original article ↗Lead image — The Decoder
Summary
TL;DR
- OpenAI cuts inference costs for guest ChatGPT users by over 50%
- Reduction in the number of Nvidia GPUs required for service
- Optimization applied primarily for unregistered users
- Potential for improved scalability and operational efficiency
- Impact on full product offerings remains unknown
Interesting facts
- Deepseek recently released an open-source method achieving 60-85% faster inference requests
- The optimization could provide resources for scaling services rather than directly impacting chip demand