🔍 Read the full analysis: Could GPT-6.1 Terra Be The Value Option In OpenAI’s Model Lineup? on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
OpenAI released GPT-6.1 Terra on October 26, 2026, positioning it between Luna and its larger Sol and Astra models. Its API prices are $1 per million input tokens and $5 per million output tokens, but independent tests show that cost per completed task depends on the effort setting and benchmark.
OpenAI released GPT-6.1 Terra on October 26, describing it as a fast, efficient model for routine work and pricing its API at $1 per million input tokens and $5 per million output tokens. Independent testing by Artificial Analysis found that Terra’s cost and performance vary by reasoning setting: its low and medium settings were on the evaluator’s cost-efficiency frontier, while higher settings cost more without matching that efficiency.
Terra is available through the API under the model name gpt-6.1-terra, as well as in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, according to OpenAI. The company said the same allowance in ChatGPT Work can cover “substantially more” tasks with Terra than with Sol or Astra. Its listed API prices are half those of GPT-5.6 Terra: $1 input and $5 output per million tokens, with cached input priced at $0.05 per million tokens.
Artificial Analysis gave Terra a score of 48.1 on its Intelligence Index at maximum effort, with an average cost of $0.40 per task. In the same comparison, GPT-6.1 Sol scored 51.8 at $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. The index combines 10 evaluations and accounts for the tokens used to complete them; its task-cost figures are averages across those evaluations, not a price for every real-world task.
The evaluator measured six effort settings, with scores ranging from 33.5 to 48.1 and average task costs from $0.055 at low effort to $0.40 at maximum effort. Artificial Analysis said low and medium effort settings sat on its cost-efficiency frontier. At the high end, maximum effort cost 7.4 times as much as low effort for about 12 additional index points. The source’s figures also show that non-reasoning mode was not the cheapest setting: it averaged $0.107 per task, compared with $0.055 at low effort.
OpenAI reported additional results from its own tests. On Agents’ Last Exam, it said Terra at medium effort scored 57.6%, compared with 42.6% for GPT-5.6 Terra at the same setting, while costing $0.29 rather than $0.84 per task. On Terminal-Bench 4.0, OpenAI said Terra at maximum effort solved 50.3% of tasks, compared with 21.5% for GPT-5.6 Terra. These are OpenAI’s reported benchmark results; the company said Sol and Astra remain better choices for the most complex coding work.
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 Depend on Effort
Terra’s launch puts a lower-priced option into OpenAI’s lineup, but its token rates alone do not show how much a task will cost. Task cost depends on token use as well as the price per token. Artificial Analysis found that Luna’s token price was lower, yet Terra at low effort cost less per evaluated task in its comparison because Terra used fewer tokens. That finding applies to the evaluator’s test tasks; it does not establish that Terra will be cheaper for every user or workload.
For businesses processing large volumes of routine work, such as invoice reading and data entry, the difference between a low-effort and maximum-effort configuration could affect operating costs. OpenAI says Terra is intended for this kind of work. The evaluation suggests that organizations should compare the price and results for their own tasks, rather than assume the lowest token rate or the strongest setting will offer the best value.
Terra’s scores also mark a trade-off within OpenAI’s range. At maximum effort it scored below Sol and several other tested models, including Claude Opus 5.5, while costing less per task in Artificial Analysis’s index. The release therefore adds a middle option, not evidence that a less expensive model can replace larger models across difficult work.
Products Worth Considering
Dell XPS 13 9350 AI Professional Laptop 13.4 120Hz FHD+ Display
FLEXISPOT 32-Inch Electric Standing Desk Converter with Keyboard Tray
Apple iPad Air 11-inch (M2), 512GB, Wi-Fi, Space Gray
As an affiliate, we earn on qualifying purchases.
Where Terra Fits in OpenAI’s Range
OpenAI presents Terra as a model between GPT-6 Luna and the larger GPT-6.1 Sol and Astra models. The company’s launch description calls it “a fast, efficient model” for communication and computer-based work. Its example is reading incoming invoices and entering details into accounting software.
Artificial Analysis’s launch-day comparison places Terra below Sol, Astra and Claude Opus 5.5 on its Intelligence Index, but above Luna. It also shows that lower-cost alternatives exist: the open-weight MiMo-V2.6-Pro scored 46.3 with an average task cost of $0.13 in the same dataset, compared with Terra’s 48.1 and $0.40 at maximum effort. These results are specific to the evaluator’s methods and benchmark set.
OpenAI also reported that Terra’s error rate on its internal set of error-inducing prompts fell from 53.6% to 16.7% at low effort and from 22.3% to 4.3% at maximum effort, compared with GPT-5.6 Terra. Artificial Analysis separately reported that Terra guessed on 38% of questions it could not answer in its hallucination test, down from 88% for GPT-5.6 Terra. The tests use different methods, so their figures should not be treated as directly comparable.
Benchmark Limits and Open Questions
Artificial Analysis’s task costs and scores summarize a defined set of evaluations; they do not predict the bill or accuracy for every customer’s workflow. Results can change with task complexity, prompt design, output length, caching and effort setting. The supplied source material does not include the full results for every one of Terra’s six settings, so the complete score-and-cost curve cannot be verified here.
OpenAI’s computer-use, coding, speed and internal error results are company-reported. The material provides no independent speed measurement for Terra, and it does not establish how the model performs across a broad range of production workloads. OpenAI’s statement that Terra meets its “Critical” cybersecurity threshold is also the company’s own classification under its Preparedness Framework.
It remains unclear how Terra’s costs and performance will compare in users’ real-world deployments, or how much of the reported savings will carry over to different task mixes. The release information also does not specify a single recommended effort setting for all use cases.
Testing Terra in Real Workflows
Terra is available now in the API, ChatGPT Work and Codex for the eligible plans named by OpenAI. Companies considering it can compare low and medium effort against their current model using representative tasks, tracking both cost per completed task and task quality. For coding or other demanding work, OpenAI says Sol and Astra remain its preferred options.
Further independent evaluations and deployment data may clarify whether Terra’s cost advantage holds outside benchmark tasks, and which effort settings offer the best balance for different workloads. Until then, the available evidence supports a narrower conclusion: Terra has lower listed token prices than its predecessor and can be cost-efficient on some tested tasks, but its value depends on the task, setting and outcome required.
Key Questions
How much does GPT-6.1 Terra cost?
OpenAI lists API prices of $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Actual task costs depend on token use and the selected effort setting.
Is Terra cheaper than GPT-6.1 Sol?
Its listed token prices are lower. In Artificial Analysis’s index, Terra at maximum effort averaged $0.40 per task, compared with $0.72 for Sol. That is an average across the evaluator’s tasks, not a guarantee for other workloads.
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 a particular user will depend on the task’s difficulty and the quality needed; maximum effort scored higher but cost more in the test.
Can Terra replace OpenAI’s larger models?
The supplied evidence does not show that. Terra scored below Sol and Astra on some reported comparisons, and OpenAI said Sol and Astra remain better choices for its most complex coding tasks.
Source: ThorstenMeyerAI.com
