OpenAI released GPT-6.1 Terra on Monday: a mid-tier model built for screen-reading, form-filling computer work — the boring half of what agents actually do.
Terra replaces GPT-5.6 Terra and closes a gap OpenAI created itself. At DevDay on 29 September, Sol and Luna moved into the GPT-6 generation; Terra stayed behind on the older release. The new model sits between Luna and Sol/Astra in the lineup, positioned on cost rather than capability. Astra remains the flagship.
The pricing is the argument. Terra costs $1 per million input tokens, 5 cents per million cached input tokens and $5 per million output tokens — 50 percent below both GPT-5.6 Terra and GPT-6.1 Sol, according to OpenAI. For long, multi-step workflows the per-token price is not the number that matters; cost per finished task is. That is the number OpenAI is selling.
The figures it published, all from its own evaluations: on Agents' Last Exam, a benchmark of professional tasks such as turning architectural drawings into a 3D model, Terra at medium reasoning effort scores 15 percentage points above GPT-5.6 Terra at roughly a third of the cost, and matches GPT-6 Astra at a tenth of its price per token. At maximum effort, OpenAI says Terra is more than three times as token-efficient as Anthropic's Claude Opus 5.5 at a quarter of the token price — roughly nine times lower cost per task. It also finishes tasks more than three times faster than Astra and GPT-6.1 Sol.
Accuracy gets a similar treatment. On an internal benchmark of prompts designed to provoke errors, the share of answers containing a factual mistake at low reasoning effort fell from 53.6 percent with GPT-5.6 Terra to 16.7 percent with 6.1 Terra, and to 4.3 percent at maximum effort — close to Astra and Sol. On Terminal-Bench 4.0, which tests command-line coding agents, Terra solved 50.3 percent of tasks at maximum effort, more than double the 21.5 percent published for GPT-5.6 Terra. For the hardest coding work, OpenAI concedes, Sol and Astra remain the better choice.
On safety, Terra reaches the “Critical” cybersecurity threshold under OpenAI's Preparedness Framework and inherits the protections built for GPT-6 Astra, with details in an addendum to Astra's system card. The company also claims strong resistance to prompt injection.
The read: this is a price-and-throughput release, not a capability jump. OpenAI is not selling the smartest model; it is selling the cheapest one that will sit at a screen long enough to finish the job — and it names a competitor while doing so. Every benchmark quoted comes from the vendor. Independent evaluations are still pending, and the track record of vendor-run agent benchmarks argues for patience.
Terra is available now in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and to developers through the API as gpt-6.1-terra. OpenAI says the same ChatGPT usage allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with Sol or Astra.
