OpenAI's GPT-6.1 Terra cuts AI agent costs in half and runs three times faster

OpenAI on Sunday launched GPT-6.1 Terra, a mid-tier model designed for computer use and visual reasoning at a fraction of the cost of its premium counterparts. At $1 per million input tokens and $5 per million output tokens, Terra costs half as much as GPT-6.1 Sol, the company’s previous mid-tier benchmark. Cached inputs drop to just $0.05 per million tokens.

The pricing cut matters because GPT-6.1 Sol was already considered competitively priced. Terra does not simply slash price at the expense of capability: OpenAI says it completes tasks more than three times faster than both GPT-6 Astra and GPT-6.1 Sol. On Terminal-Bench 4.0, a standard agentic task suite, Terra solved 50.3% of challenges at maximum reasoning effort, compared with 21.5% for the previous GPT-5.6 Terra. At medium effort, an independent test put its pass rate at 57.6% for about $0.29 per task, against $0.84 for the older model.

OpenAI positions Terra as the right tool for continuous integration pipelines, screen-automation loops, and any workload where agents run in parallel at high volume. The model supports a 1.05 million-token context window with up to 128,000 tokens of output per request, making it suitable for long agentic chains that must hold extensive state.

Independent analyst firm Artificial Analysis describes Terra as “the first Terra-class model since GPT-5.6,” adding that it delivers more intelligence for less than a third of the cost of its predecessor. Not everyone is fully sold: researcher Max Woolf characterized the model as broadly comparable to GPT-6.1 Sol, suggesting the speed and cost improvements matter more than raw benchmark gains.

GPT-6.1 Terra is available today through ChatGPT Work and Codex for Plus, Pro, Enterprise, and Edu subscribers, and via the API under the model ID gpt-6.1-terra. Codex CLI users must be on version 0.159.0 or later. As reported by BleepingComputer, Terra’s launch comes as AI infrastructure teams are weighing the cost-per-action economics of running large fleets of AI agents for automated software testing, data extraction, and UI automation at scale.