OpenAI has introduced ChatGPT Work, an enterprise AI agent that integrates ChatGPT, Codex, and GPT-5.6 to automate workplace tasks. The launch coincides with the broader rollout of GPT-5.6 models, which promise enhanced performance at reduced costs.
TL;DR
- OpenAI launches ChatGPT Work to automate enterprise workflows using GPT-5.6, ChatGPT, and Codex.
- GPT-5.6 models offer improved performance and lower operating costs, targeting enterprise scalability.
- ChatGPT Work integrates with popular business applications and provides tiered pricing for different workloads.
What happened
OpenAI has launched ChatGPT Work, an enterprise AI agent designed to automate workplace tasks by combining ChatGPT, Codex, and GPT-5.6. The platform can operate across applications, execute long-running tasks, and produce business documents, presentations, spreadsheets, and websites.
GPT-5.6 models are now generally available through ChatGPT, Codex, and the OpenAI API. These models deliver stronger performance in coding, enterprise knowledge work, cybersecurity, and scientific research while lowering inference costs and token consumption.
OpenAI has introduced tiered models: Sol for complex reasoning, Terra for mainstream enterprise applications, and Luna for lower-cost, high-volume deployments. Pricing starts at $5 per million input tokens and $30 per million output tokens for Sol, with progressively lower-cost options for Terra and Luna.
Why it matters
ChatGPT Work shifts OpenAI's enterprise strategy from benchmark leadership to performance per dollar, addressing the growing concern of inference costs for enterprises deploying AI at scale.
The launch reflects a broader trend in enterprise AI, where organizations are moving beyond experimentation and focusing on integrating AI into production workloads with cost-efficient solutions.
The tiered approach of GPT-5.6 models provides flexibility for enterprises of different sizes to optimize their AI deployments according to specific needs and budgets.
Key facts
- ChatGPT Work combines GPT-5.6, ChatGPT, and Codex to automate enterprise tasks.
- GPT-5.6 models are generally available through ChatGPT, Codex, and the OpenAI API.
- Pricing for GPT-5.6 models starts at $5 per million input tokens and $30 per million output tokens for Sol.
- GPT-5.6 Sol scored 53.6 on Agents' Last Exam, outperforming competing models with lower compute costs.
- GPT-5.6 Sol achieved a score of 80 on the Artificial Analysis Coding Agent Index, consuming fewer than half the output tokens of competing models.
- GPT-5.6 Sol scored 73.5% on ExploitBench, up from 47.9% for GPT-5.5, and nearly doubled its predecessor's performance on ExploitGym.
- ChatGPT Work integrates with Microsoft 365, Google Drive, Slack, and Notion for enhanced productivity.
- OpenAI has incorporated robust safeguards in GPT-5.6, including model-level protections, real-time monitoring, and extensive safety testing.
Context
The launch of ChatGPT Work comes as enterprises increasingly focus on integrating AI into their production workflows, making cost efficiency a critical factor in AI adoption.
The tiered pricing model for GPT-5.6 reflects a broader industry trend towards flexible AI solutions that cater to different enterprise needs and budgets.
The emphasis on performance per dollar highlights the growing importance of cost-effective AI solutions in the enterprise market.
