AI CURATED HOT STORY · 10/27/2026
OpenAI has officially launched GPT-6.1 Terra, a new mid-tier model that significantly enhances visual reasoning and Computer Use capabilities while achieving extreme cost-efficiency. Data from Artificial Analysis indicates a 6-13 point increase in intelligence index over its predecessor, yet the cost per task has dropped to less than one-third, establishing a new Pareto optimal benchmark.
Cross-verified reports show that API input/output prices are halved, cache read costs are reduced by 75%, and it supports a massive 1.05M context window. In key benchmarks like Agents' Last Exam, it scores approximately 60%, matching top-tier model GPT-6 Astra but at one-tenth the token price, with task execution speed increasing by over 3x. It also approaches flagship Sol performance on coding agent benchmarks like Terminal-Bench 4.0.
This release marks the beginning of an era for low-cost, high-throughput AI automation workflows. Now available on ChatGPT Work, Codex, and OpenRouter, GPT-6.1 Terra is ideal for routine cross-application tasks such as invoice processing and form filling. For developers, it offers near-flagship agentic capabilities at a minimal barrier, likely accelerating the large-scale deployment and ecosystem adoption of enterprise AI agents.