🔍 Read the full analysis: GPT-6.1 Terra Makes A Case For Affordable, Capable AI on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
OpenAI released GPT-6.1 Terra on October 26, 2026, pricing API use at $1 per million input tokens and $5 per million output tokens. Artificial Analysis found its low- and medium-effort settings on the cost-efficiency frontier in its benchmark, but Terra scores below several more capable models, and higher effort costs more without matching the best value.
OpenAI released GPT-6.1 Terra on October 26, positioning it as a lower-cost model for routine workplace and computer-based tasks rather than as its most capable system. API pricing starts at $1 per million input tokens and $5 per million output tokens; an independent Artificial Analysis evaluation found Terra’s low- and medium-effort settings offered comparatively strong benchmark performance for their measured task costs.
Terra is available in the API under the model name gpt-6.1-terra and to eligible Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. OpenAI describes it as a fast, efficient option for work such as reading invoices and entering data into accounting software. The listed API rates are half the rates given for GPT-5.6 Terra and GPT-6.1 Sol; cached input costs $0.05 per million tokens.
Artificial Analysis reported a score of 48.1 on its Intelligence Index for Terra at maximum effort, at an average cost of $0.40 per task across its evaluation set. Its figures put GPT-6.1 Sol at 51.8 and $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. These are benchmark results, not guarantees about the cost or quality of a particular customer’s workload. Artificial Analysis said Terra’s low and medium settings sat on its cost-efficiency frontier; settings above those did not.
OpenAI’s launch-page data also reports gains on selected tests. At medium effort, Terra scored 57.6% on Agents’ Last Exam for computer use, compared with 42.6% for GPT-5.6 Terra, at stated costs of $0.29 and $0.84 per task. On Terminal-Bench 4.0, OpenAI reports a 50.3% score at maximum effort, against 21.5% for the previous model’s published result. OpenAI says its internal factual-error rate fell at both low and maximum effort, but those findings come from the company’s own evaluations.
Infographic 01 · The effort curve
One model, six settings, a 7-fold cost range.
GPT-6.1 Terra at each effort level, as measured by Artificial Analysis. Low and medium sit on its cost-efficiency frontier: no tested model is both cheaper and better.
Score · Intelligence Index
independent higher is better
Cost per finished task
note the last step shorter is cheaper
All six settings
| Effort | Index | Cost / task | Output tokens / task | Hallucination rate |
|---|---|---|---|---|
| Non-reasoning | Index 33.5 | Cost $0.107 | Tokens 4,737 | Hallucination 72% |
| Low | Index 36.2 | Cost $0.055 | Tokens 2,681 | Hallucination 46% |
| Medium | Index 41.3 | Cost $0.095 | Tokens 5,835 | Hallucination 44% |
| High | Index 43.9 | Cost $0.134 | Tokens 9,027 | Hallucination 43% |
| Xhigh | Index 45.4 | Cost $0.194 | Tokens 14,551 | Hallucination 42% |
| Max | Index 48.1 | Cost $0.403 | Tokens 35,096 | Hallucination 38% |
Hallucination rate in AA-Omniscience: how often the model answers anyway when it does not know. Lower is better.
The first steps up are cheap. The last one doubles the bill.Assessment: Thorsten Meyer
Sources: Artificial Analysis model pages (retrieved 26 Oct 2026); Artificial Analysis launch article (26 Oct 2026). Cost per task is the average across the index’s evaluation tasks.
Powered by Thorsten Meyer AI
Task Costs Shape Terra’s Value
Terra’s launch highlights why price per token is not the same as cost per completed task. A cheaper model may use more tokens or need repeated attempts, while a more expensive model may finish a task with fewer calls. Artificial Analysis found that Terra’s low-effort setting cost less per benchmark task than GPT-6 Luna, despite Luna’s lower listed token rates. The result reflects the benchmark’s measured workload and should not be treated as a universal price comparison.
For organizations handling large volumes of routine work, a lower cost per successful task could make automation more economical, provided accuracy and oversight meet their needs. OpenAI says Terra’s user allowances cover substantially more tasks than with Sol or Astra, but that is the company’s product claim. The benchmark figures also show a trade-off: Terra does not match the highest-scoring models, and OpenAI itself recommends Sol or Astra for the most complex coding tasks.
The practical choice is likely to depend on task difficulty. Low or medium effort may suit simpler, repeatable work; higher-effort reasoning may be justified where quality matters more than expense. Users and companies will need to test their own workloads rather than rely on a single headline price or benchmark score.
Popular right now
Ninja SP151 Flip Toaster Oven and Air Fryer, 8-in-1, Stainless Steel
Ninja Foodi 10-Quart DualZone 6-in-1 Air Fryer, Black, DZ302
Roborock Q7 M5+ Robot Vacuum and Mop
As an affiliate, we earn on qualifying purchases.
Where Terra Fits in OpenAI’s Lineup
OpenAI places Terra between GPT-6 Luna, its smaller model, and the more capable Sol and Astra models. Artificial Analysis measured Terra at 48.1 at maximum effort, compared with 38.1 for Luna and 51.8 for Sol. It also rated several other systems above Terra, including Claude Opus 5.5 and Kimi K3.1 Pro, while Xiaomi’s open-weight MiMo-V2.6-Pro scored 46.3 at a lower measured task cost.
Terra offers six effort settings, from non-reasoning to maximum. In the supplied Artificial Analysis figures, maximum effort costs $0.40 per task and produces a score of 48.1; the source says the full curve ranges from 33 to 48 points and $0.05 to $0.40 per task. The source also reports that maximum effort costs 7.4 times as much as low effort for 12 additional points. Notably, non-reasoning is not the cheapest setting: it costs almost twice as much per task as low effort while scoring lower.
Artificial Analysis’s Intelligence Index combines ten evaluations, including agentic office work, terminal tasks, science and hallucination. Its task-cost figure is an average across those tests and includes token use associated with cached inputs and reasoning. The publisher said its results were retrieved on October 26, 2026; benchmark rankings and provider prices can change as models and rates are updated.
Real-World Costs Still Need Testing
The benchmark results do not establish how Terra will perform across every company’s workflows. Actual task costs and error rates can vary with prompts, task complexity, input length, retries and the amount of human review required. Artificial Analysis had not published speed measurements for Terra at the time covered by the source, despite OpenAI’s claim that Terra completes tasks more than three times faster than Astra and Sol.
Some of the most favorable performance claims come from OpenAI’s own tests, including its computer-use, factual-error and coding figures. The company also said Terra meets the Critical threshold for cybersecurity under its Preparedness Framework and retains safeguards developed for GPT-6 Astra; that classification is OpenAI’s assessment, not an independent finding described in the source. The supplied material does not give full details of the six-setting table, including every score, cost and hallucination-rate value, so those figures cannot be assessed here.
It is also unclear how the benchmark’s average task costs compare with the all-in expense of production use, including monitoring, tool access and staff review. No customer results or independently verified speed data were provided in the source material.
Measure Results on Actual Workloads
Terra is available now through the listed API and OpenAI products. The next useful test for businesses will be to run comparable tasks across Terra’s effort settings and alternative models, recording cost per successfully completed task, accuracy, speed and the need for human corrections. For coding or other demanding work, OpenAI’s own guidance points customers toward Sol or Astra.
Further independent measurements, especially on speed and real-world deployments, could clarify whether Terra’s benchmark cost advantage carries over to production. The source does not identify a future release date or additional evaluation schedule, so the timing of new results remains unknown.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model aimed at routine workplace and computer-based tasks. It is available through the API as gpt-6.1-terra and in ChatGPT Work and Codex for eligible plans.
How much does GPT-6.1 Terra cost?
OpenAI’s listed API price is $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Those rates do not by themselves show the cost of completing a task.
Is Terra OpenAI’s most capable model?
No. Artificial Analysis scored Terra at 48.1 at maximum effort, below GPT-6.1 Sol at 51.8 and several other evaluated models. OpenAI says Sol and Astra are better choices for its most complex coding tasks.
Which Terra effort setting is the best value?
Artificial Analysis found that low and medium effort sat on its cost-efficiency frontier. The best setting for an individual task depends on the required accuracy and complexity; the benchmark does not establish one setting as best for every workload.
Source: ThorstenMeyerAI.com
