OpenAI Launches GPT-6.1 Terra, Halving API Price for Agent Work

OpenAI has officially introduced GPT-6.1 Terra, a high-speed computer-use and vision model designed specifically to enhance agent workloads. The new release aims to make automated AI agent operations significantly more affordable and accessible for developers and enterprises building complex workflows.

The model is priced at 1 dollar for input tokens and 5 dollars for output tokens per million tokens. This strategic pricing effectively halves the API cost of the previous GPT-6.1 Sol model for running intensive agent-based operations that require continuous reasoning and tool execution.

In terms of benchmarks, GPT-6.1 Terra achieves the same medium-effort score on the Agents’ Last Exam as the more powerful GPT-6 Astra model. However, it delivers this performance at a fraction of the cost, dropping the expense down to just 0.29 dollars per task.

This sharp reduction from the previous 4.82 dollars per task milestone represents a major turning point for commercial AI deployment. Lowering operational expenses allows companies to scale up autonomous agent systems without encountering prohibitive API overhead costs.

Stock photo for illustration only, not from the actual event

Source: MarkTechPost