🔍 Read the full analysis: GPT-6.1 Terra: A Cost-Conscious Alternative For OpenAI Users 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 launched 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 cost per task ranges from $0.05 at low effort to $0.40 at maximum effort, with higher settings gaining benchmark points but not always better value. Some performance and safety claims come from OpenAI’s own tests, and results depend on the benchmark and task.
OpenAI describes Terra as a fast, efficient model for communication and computer-based work, including high-volume tasks such as reading invoices and entering information into accounting software. The company made it available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, as well as through the API under the model name gpt-6.1-terra. OpenAI says eligible ChatGPT users can complete substantially more tasks within the same allowance than with Sol or Astra; that is the company’s characterization, not an independently verified allowance comparison.
Artificial Analysis’s launch-day evaluation gives Terra a 48.1 Intelligence Index score at maximum effort, at an average cost of $0.40 per task across its evaluation tasks. Its six-setting results show a trade-off: low effort scored 36.2 at $0.055 per task, while maximum effort scored 48.1 at $0.40. The firm says low and medium settings are on its cost-efficiency frontier; higher settings cost more without offering the best score-for-cost trade-off among tested models. The index combines ten evaluations, and its task-cost figure is an average across those tasks, not a price guaranteed for every user request.
OpenAI’s own launch-page chart reports improvements on selected tests. At medium effort, Terra scored 57.6% on Agents’ Last Exam, compared with 42.6% for GPT-5.6 Terra, and OpenAI lists the cost per task as $0.29 versus $0.84. On Terminal-Bench 4.0, OpenAI reports a 50.3% score at maximum effort, compared with 21.5% for the predecessor. These figures are OpenAI-reported results; they should not be treated as direct comparisons across unrelated benchmarks or as a guarantee of performance on ordinary workloads.
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 Change the Comparison
Terra’s launch highlights why token price alone can mislead when comparing models. A cheaper rate per million tokens does not necessarily mean a cheaper completed task: output length, reasoning effort and the task itself affect the bill. Artificial Analysis found Luna’s per-token price is lower, but Terra at low effort cost less per task in its evaluation, in part because Terra generated fewer tokens. That result applies to the firm’s measured tasks, not every use case.
For businesses processing many routine requests, a lower average task cost could make it practical to use an AI model for work that would be too expensive at higher rates. But the benchmark also shows a quality-cost trade-off: raising Terra from low to maximum effort cost about 7.4 times as much for 12 additional Intelligence Index points. Teams considering deployment will need to test representative work, monitor accuracy and budget around completed tasks rather than relying solely on published token rates.
Terra is positioned below OpenAI’s larger models, not as a top-scoring substitute for every job. Artificial Analysis scored GPT-6.1 Sol at 51.8 and Claude Opus 5.5 at 57.6 at maximum effort, compared with Terra’s 48.1. OpenAI also says Sol and Astra remain better choices for the most complex coding tasks. The practical question is whether Terra’s lower cost and sufficient performance fit a particular workload.
Popular right now
Roborock Q7 M5+ Robot Vacuum and Mop
Philips 3300 Series Fully Automatic Espresso Machine with LatteGo Milk System
Ninja SP151 Flip Toaster Oven and Air Fryer, 8-in-1, Stainless Steel
As an affiliate, we earn on qualifying purchases.
How Terra Fits OpenAI’s Model Line
OpenAI places Terra between the smaller Luna and larger Sol and Astra models. Its stated role is everyday communication and computer-based work, rather than the most demanding tasks. The launch price is $1 per million input tokens and $5 per million output tokens, with cached input priced at $0.05 per million tokens, according to the source material. GPT-5.6 Terra was listed at $2 and $12 per million input and output tokens, respectively.
Artificial Analysis provides an independent comparison using its own combined benchmark and task-cost method. In the published figures supplied for this report, Terra’s maximum-effort score trails several higher-scoring models, including Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon. It also sits among several lower-cost alternatives: the open-weight MiMo-V2.6-Pro scored 46.3 for an average $0.13 per task in the same dataset. These rankings describe that index and its listed prices, rather than every model’s performance across all applications.
The source report also says Terra declined to answer more often than its predecessor in Artificial Analysis’s hallucination test: it guessed on 38% of questions it could not answer, down from 88% for GPT-5.6 Terra. OpenAI separately reported lower error rates on its internal error-inducing prompts. Because these are distinct tests, they do not establish a single general error rate for Terra.
Workloads Still Need Testing
Published scores cannot establish how Terra will perform or what it will cost on an organization’s own tasks. Artificial Analysis’s cost per task is an average across its index evaluations, while OpenAI’s computer-use, coding and factual-error figures come from company-selected tests. The source material does not provide enough detail to independently reconcile every difference between the two sets of results.
Some details also remain unavailable in the supplied reporting. Artificial Analysis had not published speed measurements for Terra at the time of the report, so OpenAI’s statement that it completes tasks more than three times faster than Sol and Astra has not been independently checked there. OpenAI’s classification of Terra as meeting the Critical threshold in cybersecurity under its Preparedness Framework is also the company’s own assessment. Real-world accuracy, latency, rate limits and total costs across specific workloads remain to be established.
Compare Results on Real Tasks
The next useful evidence will come from independent speed measurements and evaluations on workloads beyond the launch benchmarks. Organizations considering Terra can compare low and medium effort against their current model using the same prompts, completion criteria and quality checks, then track cost per successfully completed task rather than token charges alone.
OpenAI’s launch makes Terra available through ChatGPT Work, Codex and the API, so users can begin testing it within those products. The source material does not specify a future benchmark date or additional release milestone. Until more comparable results are published, the decision between Terra and larger models will depend on the complexity of the work and the accuracy and cost a user observes in practice.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model intended for communication and computer-based tasks. OpenAI positions it between Luna and its larger Sol and Astra models.
How much does GPT-6.1 Terra cost?
The listed API price is $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Artificial Analysis measured average task costs from $0.05 at low effort to $0.40 at maximum effort across its evaluation tasks.
Is Terra cheaper than GPT-6.1 Sol?
Its listed token rates are lower than Sol’s, and Artificial Analysis reported an average maximum-effort task cost of $0.40 for Terra versus $0.72 for Sol in its index. Actual costs depend on the task, token use and effort setting.
Is Terra OpenAI’s strongest model for coding?
No. OpenAI says Sol and Astra remain better choices for the most complex coding tasks. Its reported Terminal-Bench result for Terra is a benchmark score, not evidence that it is best for every coding workload.
Where can users access GPT-6.1 Terra?
OpenAI says Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and through the API as gpt-6.1-terra.
Source: ThorstenMeyerAI.com
