OpenAI launches GPT-5.2, pushing harder into agentic workflows

OpenAI launched GPT-5.2, a new flagship model family positioned for everyday professional use and multi-step “agentic” work (e.g., spreadsheets, presentations, code, tool-use, and long-context tasks). The release underscores the competitive pressure among frontier labs, with OpenAI framing GPT-5.2 as a direct response to a rapidly intensifying benchmark and product race. The rollout includes multiple variants (e.g., faster “Instant” and more deliberate “Thinking/Pro” modes), signaling continued segmentation by latency/price/reasoning depth for enterprise and power users. Strategically, GPT-5.2 strengthens OpenAI’s claim that the next battleground is dependable end-to-end task execution, not just raw benchmark scores, and it foreshadows more “assistant/agent” features being productized across ChatGPT and the API.