OpenAI releases GPT-6.1 Terra, halving API prices and adding computer use features

What happened

OpenAI has released GPT-6.1 Terra, a new mid-range artificial intelligence model built to handle everyday office tasks and automated software control (the-decoder.com). According to the company, the model combines visual reasoning, computer use capabilities, and a natural writing style for professional workloads like drafting proposals, interpreting technical drawings, and filling out forms (the-decoder.com).

The model is available now for Plus, Pro, Enterprise, and Edu users through ChatGPT Work and Codex, as well as to developers via the API under the name gpt-6.1-terra (the-decoder.com, helpnetsecurity.com). Standard API pricing for Terra is set at $1 per million input tokens and $5 per million output tokens, which is 50 percent lower than GPT-5.6 Terra and GPT-6.1 Sol (the-decoder.com, helpnetsecurity.com). Cached input tokens cost $0.05 per million (the-decoder.com).

Why it matters

Analysts and benchmarks point to significant cost-efficiency gains. Artificial Analysis notes that GPT-6.1 Terra achieves a strong price-to-performance ratio, gaining points on the Intelligence Index over GPT-5.6 Terra while operating at a lower cost per task (the-decoder.com). On professional computer task benchmarks such as Agents' Last Exam, OpenAI reports that Terra performs comparably to its top-tier GPT-6 Astra model while costing one-tenth of the token price, and finishes tasks more than three times faster than both Astra and GPT-6.1 Sol (the-decoder.com, digitaltrends.com).

In programming evaluations, Terra solved 50.3 percent of tasks on Terminal-Bench 4.0 at maximum reasoning effort, more than doubling the 21.5 percent score achieved by GPT-5.6 Terra (helpnetsecurity.com). OpenAI reports that its internal error rate for factual inaccuracies at low reasoning effort dropped from 53.6 percent to 16.7 percent (note.com). However, the company continues to recommend GPT-6.1 Sol or GPT-6 Astra for the most demanding coding and research workloads (helpnetsecurity.com, decrypt.co).

Context

GPT-6.1 Terra meets OpenAI's "Critical" cybersecurity threshold under its Preparedness Framework, inheriting protections from GPT-6 Astra (helpnetsecurity.com). According to the model's system card, Terra defended against indirect prompt injection attacks at a rate of 98.8 percent, up from 96.2 percent with GPT-5.6 Terra (the-decoder.com).

What we don't know yet

While OpenAI highlights high performance and lower token costs, external experts note that real-world business expenses may also involve additional costs such as inference tokens, tool usage fees, retries, and long-form input charges (note.com). Furthermore, independent verifications of benchmark claims and complex multi-app enterprise workflows remain to be extensively tested outside of controlled environments.