OpenAI Rolls Out Cheaper AI Models and Agent Tools to Cut Costs and Expand Mainstream Adoption
Summary
OpenAI is rolling out cheaper AI models and powerful agent tools, slashing costs by up to 80% after enterprise backlash over massive token bills, while pushing AI agents into the mainstream for its 990 million ChatGPT users.
Key Points
- OpenAI's agent harness, the underlying technology powering both Codex and ChatGPT Work, is emerging as the company's secret weapon, acting as an orchestration layer that manages user requests, model interactions, tools, and task execution.
- In response to enterprise backlash over skyrocketing token costs — some developers hitting $20,000 monthly bills — OpenAI is rolling out optimized models including GPT-5.6 Sol, which outperforms rivals while using 54% fewer output tokens, and GPT-5.5 Luna, which costs 80% less than Sol.
- OpenAI is pushing AI agents to the mainstream through ChatGPT Work, targeting the 990 million ChatGPT users who don't use Codex, with expanded mobile and cloud access, voice upgrades, and a call for users to tackle more ambitious, multi-day tasks with AI.