OpenAI releases GPT-6.1 Terra at half the API price of its predecessor

OpenAI launched GPT-6.1 Terra on October 26, 2026, a fast, efficient model aimed at everyday office work and automated computer control. Standard API pricing is $1 per million input tokens, $0.05 per million cached input and $5 per million output, half the price of GPT-5.6 Terra, and the model is live now in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, plus the API under the ID gpt-6.1-terra.

A price cut with a capability bump

Terra is OpenAI's mid-tier line, and this generation moves both price and performance. On Agents' Last Exam, a benchmark of professional computer-use tasks like turning architectural drawings into 3D models, Terra scores 15 percentage points higher than GPT-5.6 Terra at medium reasoning effort while costing about a third per run. OpenAI says its performance matches GPT-6 Astra at roughly one-tenth Astra's price per token.

Against Anthropic, the comparison is direct: at maximum effort, OpenAI claims Terra is over three times as token-efficient as Claude Opus 5.5 at a quarter of the token price, about nine times lower cost per task, and completes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol. The Decoder's coverage puts Terra's input/output prices at half of what Claude Sonnet 5.5 currently charges, framing Terra as OpenAI's closest answer to Anthropic's mid-tier Sonnet line.

Hallucinations drop sharply

The most striking numbers are on factuality. On OpenAI's internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% for GPT-5.6 Terra to 16.7% for Terra, a drop of nearly 37 percentage points. At maximum effort, errors fell 18 points to 4.3%, nearly matching GPT-6 Astra and GPT-6.1 Sol.

Third-party data points the same direction. Artificial Analysis, as reported by The Decoder, measured the hallucination-guessing rate on its AA-Omniscience benchmark falling from 88% for GPT-5.6 Terra to 38% for the new model, and placed Terra on the Pareto frontier for intelligence per dollar, scoring 48 on the Intelligence Index, between GPT-5.6 Terra (42) and GPT-6.1 Sol (52), at under a third of the per-task cost of its predecessor ($0.40 versus $1.40).

Coding gets serious at the mid tier

On Terminal-Bench 4.0, Terra solves 50.3% of tasks at maximum reasoning effort, more than double GPT-5.6 Terra's published 21.5%. That score puts it in the same tier as GPT-6.1 Sol, according to Artificial Analysis. On Terminal-Bench-Science 0.1, Terra outperforms Sol at lower cost and lands within five points of Opus 5.5 at less than a fifth of the estimated cost.

OpenAI still recommends GPT-6.1 Sol and GPT-6 Astra for the most complex coding work, so Terra is a volume-and-efficiency play, not the new flagship.

Safety posture and what to verify

Terra meets the Critical threshold in cybersecurity under OpenAI's Preparedness Framework and keeps the safeguards introduced with GPT-6 Astra. The system card reports a 98.8% defense rate against indirect prompt injection, up from 96.2% for GPT-5.6 Terra. The benchmark figures above are largely vendor-reported; independent analysis from Artificial Analysis, via The Decoder, corroborates the price-performance claim, but real-world agent reliability at the half-price tier is still worth testing before re-routing production workloads.

For teams running high-volume pipelines, the practical question is whether the halved price holds at your reasoning-effort mix, and whether cached-input savings at $0.05 per million apply to your prompt shapes. Check usage-allowance math in your workspace before switching defaults.

AI model and pricing announcements like this change fast. LinkLoot tracks availability and workflow impact across providers; see the guide to AI agent tools at /guides/ai-agent-tools for context on choosing models per task.