OpenAI Launches GPT-6.1 Terra: A Faster, Cheaper AI Model for Everyday Office Work

OpenAI Launches GPT-6.1 Terra: Efficiency and Lower Costs for Everyday Work

OpenAI announced yesterday (Oct 26) GPT-6.1 Terra, a new AI model that combines precise visual reasoning, strong computer-use capability, and clearer, more natural writing. The model is built specifically for the kind of tasks that fill a workday: drafting proposals, interpreting technical drawings, filling out forms, editing videos, and reading incoming invoices to enter their details into accounting software.

What makes this especially relevant for small and medium businesses is the pricing: standard API rates for Terra are 50% lower than the previous version (GPT-5.6 Terra) and GPT-6.1 Sol, making high-volume automation - like invoice processing - significantly more economical. At the same time, factual error rates dropped sharply: from 53.6% to 16.7% at low reasoning effort, and down to just 4.3% at maximum effort - nearly matching the company's more expensive flagship models.

The model is available today to all Plus, Pro, Enterprise, and Edu users within ChatGPT Work and Codex, and to developers via the API as gpt-6.1-terra.

Source: https://openai.com/index/introducing-gpt-6-1-terra/

Small Retailers vs. AI Shopping Agents

Meanwhile, The RealReal, a luxury secondhand fashion marketplace, had to tighten its bot defenses after it was revealed that Meta's Muse AI agent could exploit the site's 20-minute item-hold system by repeatedly re-adding items to a cart - effectively blocking other shoppers from buying the same products for hours. The company updated its bot protections to make it much harder for AI agents to log in, add items to cart, and check out.

This joins a broader trend: Amazon already blocks AI shopping agents from its site, worried about the impact on ad revenue and direct customer relationships. For small and medium businesses selling online, it's a practical reminder to examine now how your inventory, ordering, and reserve systems will hold up once customers start shopping through AI agents instead of browsing and choosing products themselves.

Source: https://www.theverge.com/news/1018713/meta-muse-agentic-shopping-ai-the-realreal

Briefly: OpenAI and Scientific Research

OpenAI also announced three new "discoveries" in mathematics, computer science, and physics produced using its models - this time accompanied by external academic expert groups who will review the results before full publication, an attempt to strike a more responsible balance between speed of release and scientific verification.

Source: https://openai.com/index/building-scientific-understanding-with-academic-communities/