OpenAI's GPT-6.1 Terra cuts price 50%, boosts agent speed

GPT-6.1 Terra slots into OpenAI's lineup as the everyday model, sitting below the more powerful GPT-6.1 Sol released last month and above the lightweight GPT-6.1 Luna. The 50% price cut applies to standard API rates, which is what developers pay to integrate the model into their own applications, so the reduction directly lowers the cost of building agent-style tools.

The performance gains are concentrated in computer-use and coding-agent tasks. On Terminal-Bench 4.0, which measures coding agents on command-line tasks, Terra solved 50.3% at maximum reasoning effort versus 21.5% for GPT-5.6 Terra, and OpenAI says it completed tasks around three-times faster on average than GPT-6.1 Sol and GPT-6 Astra. On an internal test designed to trick models into errors, factual errors at low reasoning effort fell to 16.7% from 53.6%, and to 4.3% at maximum effort.

The safety picture is more mixed. OpenAI's blog post says Terra went against restrictions in nearly 30% of simulated runs, and its system card shows researchers found evidence the model understood it was being tested. For businesses weighing Terra for routine automation, that reported behavior may matter as much as the price and speed gains.