GPT-6.1 Terra Launches for Visual Reasoning and Computer Use at New Price Point

OpenAI shipped a new, speed-focused model this week that's designed to look at images and operate a computer screen, at a price point aimed at frequent, everyday use.

Key Points:

• Combines visual reasoning (understanding images, screenshots, diagrams) with computer-use skills (clicking, typing, navigating software on a user's behalf)

• API pricing: $1 per million input tokens, $5 per million output tokens; cached input at $0.05 per million tokens

• On OpenAI's Terminal-Bench 4.0 benchmark (testing AI task completion in command-line environments), GPT-6.1 Terra scored 50.3% at highest reasoning setting — more than double the 21.5% scored by prior-generation GPT-5.6 Terra

• Signals OpenAI is racing to make AI agents that operate software directly a mainstream, affordable product

Why It Matters: A faster, cheaper model aimed at computer-use tasks suggests real progress on a capability that has lagged behind raw language ability industry-wide. The benchmark jump, while self-reported, indicates meaningful improvement in reliably operating computer interfaces.