🔍 Read the full analysis: How GPT-6.1 Terra Could Make OpenAI Work More Cost-Effectively on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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, pricing API use at $1 per million input tokens and $5 per million output tokens—half the listed rates for GPT-5.6 Terra. Artificial Analysis found that Terra’s low- and medium-effort settings sit on its cost-efficiency frontier, but higher settings cost more without matching that efficiency, and stronger models score better on its Intelligence Index.
OpenAI released GPT-6.1 Terra on October 26, pricing it at $1 per million input tokens and $5 per million output tokens—half the listed rates for its GPT-5.6 Terra predecessor. Independent benchmarking firm Artificial Analysis found that Terra’s low- and medium-effort settings offer the strongest cost efficiency among the settings it tested, making the model a possible lower-cost option for routine work, though it does not lead the overall performance rankings.
Terra is available in the API, ChatGPT Work and Codex. OpenAI describes it as a fast, efficient model for communication and computer-based tasks, including processing invoices and entering their details into accounting software. The company says eligible ChatGPT Plus, Pro, Enterprise and Edu users can complete substantially more tasks with the same allowance than with Sol or Astra. Those availability and allowance descriptions come from OpenAI.
Artificial Analysis measured Terra’s Intelligence Index score at 48.1 at maximum effort, with an average cost of $0.40 per task across the index’s evaluation tasks. GPT-6.1 Sol scored 51.8 at $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. The benchmark firm’s index combines ten evaluations and reports its average task cost; it is not a price quote for every real-world use.
Terra’s six effort settings show why token prices alone do not determine a task’s cost. Artificial Analysis recorded scores from 33.5 in non-reasoning mode to 48.1 at maximum effort. Low effort cost an average of $0.055 per task, while non-reasoning cost $0.107 despite its lower score of 33.5. The supplied figures show maximum effort cost $0.40 for 12 more index points than low effort, a trade-off that may not pay off for routine tasks.
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 Matter More Than Token Rates
For businesses running large volumes of repetitive work, the relevant budget may be the cost of completing a task, not simply the published rate per million tokens. A model can use fewer tokens to finish a job and cost less overall, even if its token rate is higher than a smaller model’s. Artificial Analysis found that Terra at low effort cost less per task than GPT-6 Luna in its evaluation, despite Luna’s lower listed token price.
That result supports testing models against a company’s own workloads and tracking both completion quality and cost per finished task. Terra’s low and medium settings could suit simpler, high-volume jobs; more demanding work may call for a higher effort setting or a more capable model. Benchmark averages do not establish how much any particular company will save, and real costs will vary with task design, token use and the required accuracy.
OpenAI also reported improvements on its own evaluations. It said Terra’s medium-effort score on Agents’ Last Exam was 57.6%, compared with 42.6% for GPT-5.6 Terra, at $0.29 rather than $0.84 per task. On OpenAI’s internal error-inducing prompts, the company reported a low-effort error rate of 16.7%, down from 53.6%. These are company-reported test results, not independent confirmation of performance across customers’ workflows.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Terra Fits OpenAI’s Model Line
OpenAI positions Terra between its smaller Luna model and larger Sol and Astra models. Its API price of $1/$5 per million input/output tokens is half GPT-5.6 Terra’s listed $2/$12 rates; cached input is listed at $0.05 per million tokens. Prices and product descriptions are launch-day details and may change.
Artificial Analysis’s launch-day comparison shows Terra is not the highest-scoring model in its index. Claude Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon all scored above it at the settings listed. The firm also reported that open-weight MiMo-V2.6-Pro scored 46.3 at an average $0.13 per task, so Terra’s maximum-effort setting was neither the cheapest nor the highest-scoring option in that comparison.
OpenAI says Terra completes tasks more than three times faster than Astra and Sol, but Artificial Analysis had not published speed measurements for Terra when the source material was checked on October 26. OpenAI also said Terra met the “Critical” threshold in its cybersecurity Preparedness Framework and retained safeguards developed for GPT-6 Astra. Those are OpenAI’s own classifications and statements.
Real-World Savings Remain Unverified
The available figures do not establish how Terra will perform across customers’ actual workloads or how much money organizations will save. Artificial Analysis’s figures are averages across its evaluation tasks, while OpenAI’s Agents’ Last Exam, coding and error-rate results are company-reported measurements. The source material does not provide independent replication of those OpenAI tests.
Speed comparisons also remain unsettled: OpenAI claims Terra is more than three times faster than Sol and Astra, but Artificial Analysis had not released speed measurements at the time of its launch-day review. The benchmark data supplied here list six effort settings, but do not include the full results for every setting. It is also unclear how frequently users will need higher effort levels, where Terra’s task costs rise.
Measure Terra on Real Workloads
Businesses considering Terra can compare it with their current model using representative tasks, recording cost per completed job, accuracy and review time at low and medium effort before scaling use. More complex work may warrant comparisons with Sol, Astra or other models, rather than assuming one system is best for every task.
Further independent benchmark results, including speed measurements and fuller setting-by-setting comparisons, could clarify Terra’s position. OpenAI’s launch makes Terra available through its API and named products now; the evidence provided does not specify a later release date or a future pricing change.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a model OpenAI released on October 26, 2026. The company describes it as a fast, efficient option for communication and computer-based tasks.
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. Artificial Analysis measured average task costs that varied by effort setting, from $0.055 at low effort to $0.40 at maximum effort in the supplied data.
Which Terra effort setting is most cost-efficient?
Artificial Analysis found that low and medium effort sit on its cost-efficiency frontier. Low effort averaged $0.055 per task in the supplied results; the best setting for a real workflow depends on its accuracy and complexity requirements.
Does Terra outperform OpenAI’s larger models?
Not on the supplied Artificial Analysis Intelligence Index results: Terra scored 48.1 at maximum effort, below GPT-6.1 Sol at 51.8 and GPT-6 Astra at 52.7. Terra’s lower price may make it suitable for some tasks, but the scores do not establish that it is the better choice for every workload.
Are the reported savings confirmed for businesses?
No company-wide or customer savings are confirmed in the source material. The reported figures are benchmark averages or OpenAI’s own test results; actual costs and outcomes depend on the tasks, effort settings and amount of human review.
Source: ThorstenMeyerAI.com
