🔍 Read the full analysis: GPT-6.1 Terra May Deliver The Capability You Need At A Lower Cost on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI released GPT-6.1 Terra on October 26, 2026, positioning it between its smaller Luna and larger Sol and Astra models. Artificial Analysis measured lower cost per task for Terra’s low and medium effort settings, while its highest setting scored below several leading models and cost more than low effort.
OpenAI released GPT-6.1 Terra on October 26, describing it as a fast, efficient model for everyday communication and computer-based work. Its API price is $1 per million input tokens and $5 per million output tokens, half the listed per-token price of GPT-5.6 Terra; independent testing by Artificial Analysis found Terra’s low and medium effort settings offered a comparatively low cost per completed benchmark task.
Terra is positioned between OpenAI’s smaller GPT-6 Luna and larger Sol and Astra models. OpenAI says the model is available in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, as well as through the API under the name gpt-6.1-terra. OpenAI says the same allowance in ChatGPT Work can cover substantially more tasks with Terra than with Sol or Astra. That is the company’s description; actual usage depends on the work being done.
Artificial Analysis’s launch-day evaluation gives Terra a score of 48.1 on its Intelligence Index at maximum effort, with an average cost of $0.40 per task across the benchmark’s evaluations. The firm lists GPT-6.1 Sol at 51.8 points and $0.72 per task, and Claude Opus 5.5 at 57.6 points and $5.98 per task. These figures describe the benchmark’s average task cost, not a universal price for every user request.
The model’s settings change both performance and cost. Artificial Analysis measured six effort levels, with scores ranging from 33.5 in non-reasoning mode to 48.1 at maximum effort, and task costs from about $0.05 to $0.40. Its reported curve places low and medium effort on its cost-efficiency frontier: among the tested options, no model was both cheaper and better at those points. The source material says maximum effort costs 7.4 times as much as low effort for 12 more points. It also reports that non-reasoning mode costs almost twice as much per task as low effort while scoring lower.
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
Why Cost Per Task Changes the Calculation
Terra’s launch highlights a practical distinction for organizations choosing AI models: token prices are not the same as task costs. A cheaper price per token does not guarantee a cheaper completed job. Artificial Analysis reports that Terra at low effort can cost less per task than Luna, despite Luna’s lower listed token price, because Terra uses fewer tokens in the tested tasks. The result may matter to teams processing large volumes of routine work, where token use, reasoning settings and successful completion all affect spending.
The trade-off is capability. Terra’s maximum-effort score remains below several higher-scoring models in Artificial Analysis’s comparison, including Claude Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon. The same source lists open-weight MiMo-V2.6-Pro at 46.3 points and $0.13 per task, cheaper than Terra at its higher effort settings while scoring above them. These rankings apply to the firm’s chosen evaluations and cost methodology; they do not establish which model will perform best for a particular company’s workload.
OpenAI’s own examples point to routine, repeated work such as reading invoices and entering data into accounting software. The company says Terra completes tasks more than three times faster than Sol and Astra, but Artificial Analysis had not published speed measurements for Terra at the time reported. For buyers, the useful takeaway is to test the model on their own work and track cost per successfully finished task, not rely on token price or a single overall score.
Products Worth Considering
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Dell UltraSharp 27 Thunderbolt Hub Monitor (U2726DER)
Dell UltraSharp U2725QE 27-Inch 4K UHD IPS Black Monitor
As an affiliate, we earn on qualifying purchases.
Terra’s Place in OpenAI’s Lineup
OpenAI describes Terra as a model for workday communication and computer-based tasks, rather than as its top-performing model. The company’s listed API prices are 50% lower than GPT-5.6 Terra’s: $1 versus $2 per million input tokens, and $5 versus $12 per million output tokens. Cached input is listed at $0.05 per million tokens. Artificial Analysis lists the same $1/$5 token rates for Terra and reports its own per-task averages, which account for token use on benchmark tasks.
OpenAI also published results from its own tests. On Agents’ Last Exam, it reports Terra at medium effort scored 57.6% and cost $0.29 per task, compared with 42.6% and $0.84 for GPT-5.6 Terra at medium effort. OpenAI says the Terra result matched GPT-6 Astra at medium effort, for which its chart lists a $4.82 task cost. On Terminal-Bench 4.0, OpenAI reports Terra solved 50.3% of tasks at maximum effort, compared with 21.5% for GPT-5.6 Terra. These are company-reported measurements, separate from Artificial Analysis’s independent index.
There are also safety and factuality claims in the launch material. OpenAI says Terra meets its Preparedness Framework’s “Critical” threshold in cybersecurity and retains safeguards developed for GPT-6 Astra. Artificial Analysis reports Terra guessed on 38% of questions it could not answer in its hallucination test, down from 88% for GPT-5.6 Terra. Those results refer to specific tests and do not establish error rates across all uses.
Benchmark Results Have Limits
The available comparisons do not show how Terra will perform or cost on every real-world workload. Artificial Analysis’s figures are averages across its evaluation tasks, and OpenAI’s results come from tests selected and reported by the company. Neither set of scores guarantees a particular cost, speed or accuracy for an individual user or business.
Some questions also remain open at launch. Artificial Analysis had not published speed measurements for Terra, so OpenAI’s claim that it is more than three times faster than Sol and Astra could not yet be compared with that independent source. The source material does not provide the full results for every effort setting, including all values in the effort-level curve, and does not specify how Terra’s performance varies across different customer workloads. The claimed cybersecurity classification is OpenAI’s own assessment under its Preparedness Framework.
Testing Terra on Real Workloads
The next useful evidence will come from independent speed measurements and further evaluations across coding, computer use and other work tasks. Businesses considering Terra can compare it with their current model using a representative set of jobs, recording completion rates, output quality, latency and total cost at different effort settings. That can show whether low or medium effort is sufficient, or whether a larger model is needed for more demanding work.
For now, the launch establishes Terra’s availability, its listed prices and the results reported by OpenAI and Artificial Analysis. It does not settle how much a particular user will save. That will depend on task mix, token consumption, effort selection and whether the model completes the work accurately enough to avoid additional review or retries.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model positioned between Luna and the larger Sol and Astra models. OpenAI describes it as suited to communication and computer-based work, and offers it in ChatGPT Work, Codex and the API.
How much does GPT-6.1 Terra cost?
OpenAI lists API pricing at $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Actual task costs vary with token use and the amount of reasoning required.
Does Terra outperform OpenAI’s larger models?
Not across the reported comparisons. Artificial Analysis scored Terra at 48.1 at maximum effort, below GPT-6.1 Sol at 51.8. OpenAI also says Sol and Astra remain better choices for its most complex coding tasks.
Which Terra effort setting is most cost-efficient?
Artificial Analysis found low and medium effort on its cost-efficiency frontier in the tested comparisons. The best setting for a specific job is not established by the benchmark alone; users need to weigh cost against quality and completion needs.
Are Terra’s speed and safety claims independently verified?
The source material attributes the speed and cybersecurity claims to OpenAI. It says Artificial Analysis had not published speed measurements for Terra at the time of reporting, so those claims were not independently corroborated by that firm in the available results.
Source: ThorstenMeyerAI.com
