OpenAI releases GPT-6.1 Terra with improved cost efficiency in Artificial Analysis evaluations

What happened

OpenAI has released GPT-6.1 Terra, which Artificial Analysis reports delivers higher intelligence at less than a third of GPT-5.6 Terra’s cost per task at max effort ($0.40 vs $1.40). Input/output pricing is $1/$5 per million tokens, compared with $2/$12 previously. Artificial Analysis reports a 6-point Intelligence Index gain at max effort and a hallucination rate decrease from 88% to 38% on AA-Omniscience, alongside a 2-point accuracy increase. Low and medium effort settings reach the Intelligence Index cost efficiency frontier.

Read next

• More changes in today’s briefingNvidia open-sources Flora with projected robot video labeling costs up to 125x lower than GPT-6 Astra

The story so far

OpenAI announced GPT-6.1 Terra, a model designed to interpret screens and act across apps, with rollout to paid ChatGPT subscribers reported underway and API and Codex access announced. OpenAI says Terra can interpret images, navigate software and write naturally. Baoyu reported that rollout covers ChatGPT Plus, Pro, Enterprise and Edu subscribers, with selection available in ChatGPT Work and Codex. API pricing is $1 per million input tokens and $5 per million output tokens, compared with GPT-5.6 Terra’s $2 and $12, according to Artificial Analysis. In its evaluations at maximum effort, Artificial Analysis reported a 6-point Intelligence Index improvement and a cost per task of $0.40, down from $1.40. On AA-Omniscience, it reported a hallucination-rate decline from 88% to 38%, alongside a 2-point accuracy increase. Terra scored 48 on its Coding Agent Index v2.0, below GPT-6.1 Sol’s 54 and above GPT-6 Luna’s 31. Baoyu reported that OpenAI claims average task completion more than three times faster than Astra and Sol, while still recommending those models for deep research and complex coding.