GPT-6.1 Terra halves OpenAI's API prices and still gains ground on its larger models
OpenAI has released GPT-6.1 Terra, cutting API pricing in half compared to GPT-5.6 Terra to deliver $1 per million input tokens, $5 per million output tokens, and $0.05 per million cached tokens. This places Terra at roughly half the rate of Anthropic's Claude Sonnet 5.5 ($2 input, $10 output). Artificial Analysis designated the model a new benchmark for intelligence per dollar, noting it gained six points on its Intelligence Index while reducing cost per task from $1.40 to $0.40.
The model demonstrates pronounced gains in technical and agentic workflows. On Terminal-Bench 4.0, Terra solved 50.3 percent of coding tasks at maximum reasoning effort, compared to 21.5 percent for GPT-5.6 Terra, while scoring 48 on the Coding Agent Index v2.0. On the Agents' Last Exam benchmark, it matched GPT-6 Astra at one-tenth the token cost and completed tasks more than three times faster. Terra also reduced blind guessing on the AA-Omniscience hallucination test from 88 percent to 38 percent.
On the security front, Terra satisfies OpenAI's "Critical" cybersecurity threshold, stopping 99.99 percent of direct prompt hierarchy attacks and 98.8 percent of indirect prompt injections. The model is currently accessible via the API as well as ChatGPT Work and Codex for paid and institutional tiers.
Continue in the app — vote & join in ➔