🔍 Read the full analysis: GPT-6.1 Terra Could Bring OpenAI AI Within A Tighter Budget on ThorstenMeyerAI.com
Get the latest gadgets 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 Oct. 26, 2026, pricing API use at $1 per million input tokens and $5 per million output tokens. Artificial Analysis measured lower cost per task than larger models, but Terra scores lower on its Intelligence Index and is not the top-performing model.
OpenAI released GPT-6.1 Terra on Oct. 26, presenting it as a lower-cost model for routine communication and computer-based tasks. Its listed API price is $1 per million input tokens and $5 per million output tokens, half the listed price of its predecessor, GPT-5.6 Terra; independent testing by Artificial Analysis found that Terra’s low- and medium-effort settings sit on its cost-efficiency frontier, while the model scores below more capable systems.
Terra is positioned between OpenAI’s smaller Luna model and its larger Sol and Astra models. OpenAI says it is intended for work such as reading invoices and entering data into accounting software. The company says eligible Plus, Pro, Enterprise and Edu users can access it in ChatGPT Work and Codex, and says the same allowance can cover substantially more Terra tasks than Sol or Astra. The API model identifier is gpt-6.1-terra.
Artificial Analysis gave Terra a score of 48.1 on its Intelligence Index at maximum effort, at an average cost of $0.40 per task across the index’s evaluations. For comparison, the source lists GPT-6.1 Sol at 51.8 and $0.72 per task, and Claude Opus 5.5 at 57.6 and $5.98 per task. Those figures do not mean Terra matches the higher-scoring models: they show a lower measured cost alongside a lower score.
The benchmark also points to the importance of effort settings. Artificial Analysis measured six settings, with scores ranging from 33.5 in non-reasoning mode to 48.1 at maximum effort and task costs from $0.055 at low effort to $0.40 at maximum. The source reports that moving from low to maximum costs 7.4 times as much for 12 additional index points. It says low and medium settings are on the cost-efficiency frontier; settings above them are not.
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
Terra’s Cost Depends on the Task
For businesses that run large volumes of routine tasks, cost per completed task may offer a more useful budgeting measure than price per token alone. A model can have a higher token price but use fewer tokens to finish a job. The source says Luna is ten times cheaper per token, while Terra at low effort costs less per task in the cited comparison because it generates fewer tokens. That is a result from the reported evaluation, not a guarantee for every workflow.
The trade-off is capability. Terra’s maximum-effort score remains below Sol and several other tested models, and the source says OpenAI itself advises that Sol and Astra remain better choices for the most complex coding tasks. For buyers, the practical question is whether Terra can complete a given job reliably at a lower total cost—not simply which model has the cheapest listed token rate.
Artificial Analysis also reported that Xiaomi’s open-weight MiMo-V2.6-Pro scored 46.3 for $0.13 per task in its comparison, lower in score than Terra at maximum effort but cheaper and higher-scoring than Terra’s high and extra-high settings, according to the source. That makes the comparison a competitive one rather than a simple case of Terra being the least expensive option.
Products Worth Considering
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How OpenAI Frames the Launch
OpenAI describes Terra as a fast, efficient model for communication and computer-based work. Its launch-page examples include computer use, coding and factual-error testing. The source reports OpenAI’s measurements, including a 57.6% result for Terra at medium effort on Agents’ Last Exam, compared with 42.6% for GPT-5.6 Terra; the reported cost per task was $0.29 and $0.84, respectively. OpenAI’s launch materials also show a 60.0% score for Terra at maximum effort on that evaluation, compared with 63.2% for Claude Opus 5.5, whose reported task cost was higher.
Those are OpenAI’s own benchmark claims, separate from Artificial Analysis’s independent index. OpenAI also reports improvements on its internal error-inducing prompts and a higher score on Terminal-Bench 4.0 than its predecessor. It says Terra completes tasks more than three times faster than Sol and Astra, but Artificial Analysis had not published speed measurements for Terra when the source checked on launch day. OpenAI’s cybersecurity classification is also the company’s own assessment under its Preparedness Framework.
Artificial Analysis published its evaluation on launch day, with figures retrieved Oct. 26, 2026. Its Intelligence Index combines ten evaluations and accounts for costs including cached input and reasoning tokens. Its results therefore provide a separate comparison, but the index’s average costs and scores may not predict performance or spending for a particular customer’s workload.
Benchmark Results Have Limits
The published figures do not establish how Terra will perform across every customer’s data, software or task mix. Artificial Analysis’s cost per task is an average across its evaluation suite, and its scores are not a direct prediction of a company’s real-world accuracy or total bill. Independent speed measurements were not available in the source’s launch-day review, so OpenAI’s claim that Terra is more than three times faster than Sol and Astra had not been independently checked there.
Some key figures also come from OpenAI’s own tests, including its internal factual-error prompts and its cybersecurity classification. The source does not provide enough detail to independently assess those internal results. Pricing and access can also vary by product and plan; the stated API rates should not be assumed to describe every ChatGPT subscription or the cost of every completed task.
Watch Real-World Costs and Access
The next useful evidence will be whether customers see similar savings on their own workloads and whether independent evaluators publish speed and reliability measurements. Businesses considering Terra can compare it with larger models on a representative set of routine tasks, tracking completion rate, token use and cost per successful result rather than relying on token prices alone. The source recommends starting with low or medium effort and routing harder tasks to more capable models, a strategy that should be tested against each organization’s requirements.
OpenAI’s announcement says Terra is available through the API and in ChatGPT Work and Codex for eligible plans. Further updates to benchmark coverage, pricing details or product availability could clarify how broadly the lower-cost positioning applies. Until then, the launch establishes Terra’s listed price and initial benchmark results, not a settled measure of savings for all users.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is an OpenAI model announced Oct. 26, 2026, 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 listed at $0.05 per million tokens in the source. Actual cost per task depends on token use and the task; Artificial Analysis measured an average of $0.055 at low effort and $0.40 at maximum effort across its evaluations.
Is Terra better than OpenAI’s larger models?
Not on the cited overall benchmark score. Artificial Analysis scored maximum-effort Terra at 48.1, below GPT-6.1 Sol at 51.8. Terra’s lower measured task cost may make it suitable for some routine work, but the source says Sol and Astra remain better choices for OpenAI’s most complex coding tasks.
Why compare cost per task instead of price per token?
Token rates do not show how many tokens a model uses to complete a job. A model with a higher per-token price can cost less for a particular task if it uses fewer tokens. The source’s comparisons are based on benchmark averages, so organizations should test their own workflows before estimating savings.
Source: ThorstenMeyerAI.com
