🔍 Read the full analysis: Could A Less-Smart GPT-6.1 Terra Be The More Practical Choice? 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 found that Terra’s low- and medium-effort settings offer strong cost efficiency on its benchmark, while higher effort costs more and still scores below leading models.
OpenAI released GPT-6.1 Terra on October 26, presenting it as a lower-cost model for everyday workplace tasks rather than its strongest system. Independent benchmarker Artificial Analysis measured Terra at 48.1 on its Intelligence Index at maximum effort, below several competing models, but found its low- and medium-effort settings on the benchmark’s cost-efficiency frontier.
Terra is available through the OpenAI API, ChatGPT Work and Codex, according to the launch material. Its listed API price is $1 per million input tokens and $5 per million output tokens, compared with $2 and $12 for its predecessor, GPT-5.6 Terra. Cached input is priced at $0.05 per million tokens. OpenAI says eligible ChatGPT Work users can complete substantially more tasks within the same allowance than with Sol or Astra.
Artificial Analysis’s launch-day evaluation put Terra at 48.1 points and $0.40 per task at maximum effort. 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. These are averages across the evaluation tasks, not a universal price for every use. Artificial Analysis described Terra’s low- and medium-effort settings as lying on its cost-efficiency frontier; the source material reports six effort levels spanning scores of 33 to 48 and costs of $0.05 to $0.40 per task.
OpenAI’s own published results cover several different tests and should be distinguished from the independent index. It reports that Terra at medium effort scored 57.6% on Agents’ Last Exam for $0.29 per task. On Terminal-Bench 4.0, it reports a 50.3% score at maximum effort, compared with 21.5% for GPT-5.6 Terra. OpenAI also says Terra completes tasks more than three times faster than Astra and Sol; Artificial Analysis had not published speed measurements for Terra at the time of its evaluation.
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
When Lower Task Costs Matter
The launch highlights a practical distinction for businesses and developers: token prices do not by themselves determine the cost of completing work. A model that generates fewer tokens or finishes with less reasoning may cost less per task even if another model has a lower listed price per token. Artificial Analysis’s results suggest Terra at low effort can be cheaper per task than the much smaller Luna, despite Luna’s lower token rates.
That can matter for high-volume workflows such as processing invoices, sorting routine requests or handling computer-based office tasks. Lower costs could let teams run more jobs within a fixed budget. But the benchmark results do not establish that Terra will be the best or cheapest option for every company’s workload: task complexity, accuracy requirements, token use and integration costs can change the comparison.
Terra also does not replace the need for stronger models on difficult work. Artificial Analysis ranked several models above it, and OpenAI says Sol and Astra remain better choices for its most complex coding tasks. The practical choice is therefore likely to depend on whether a particular task benefits more from lower cost and throughput or from higher benchmark performance.
Popular right now
Shark Rocket Pet Plus Corded Stick Vacuum, Bordeaux, HV322
Breville Bambino Espresso Machine, Brushed Stainless Steel
Breville Barista Pro Espresso Machine with Grinder, Brushed Stainless Steel
As an affiliate, we earn on qualifying purchases.
Terra Between Luna and Sol
OpenAI places Terra between Luna, its smaller model, and Sol and Astra, which sit higher in its model range. The company describes Terra as a fast, efficient model for communication and computer-based work. Its example is reading incoming invoices and entering the details into accounting software. That use case indicates the intended market, but it is not independent proof of performance in a customer’s own systems.
Artificial Analysis combines ten evaluations in its Intelligence Index, including agentic office work, terminal tasks, science and hallucination. Its reported cost per task accounts for tokens used in those evaluations, including cached input and reasoning tokens. On the firm’s comparison, Terra at maximum effort scored below Claude Opus 5.5, Kimi K3.1 Pro, GPT-6 Astra, Gemini 4 Argon and GPT-6.1 Sol. It was not the only lower-cost option: the source reports Xiaomi’s open-weight MiMo-V2.6-Pro at 46.3 points and $0.13 per task.
Effort settings change both performance and cost. Artificial Analysis measured Terra from non-reasoning through maximum effort; its reported figures show that maximum effort costs about 7.4 times as much as low effort for 12 more Intelligence Index points. The non-reasoning setting was not the cheapest in the reported figures: it cost nearly twice as much per task as low effort while scoring lower. Those results caution against assuming that turning reasoning off automatically reduces total task costs.
How Terra Performs in Real Work
The available figures come from benchmark tests and OpenAI’s own reporting; they do not show how Terra performs across every customer workflow. The launch-day material does not establish whether its task-cost advantage holds when prompts, software integrations, error checking and human review differ from the tests.
Some reported results are not directly comparable because they come from different evaluations. OpenAI’s Agents’ Last Exam and Terminal-Bench results are separate from Artificial Analysis’s Intelligence Index, and the claims about speed and internal factual-error tests are OpenAI’s own. Artificial Analysis had not published Terra speed measurements when the source report checked. The material also does not provide enough detail to assess the reliability of every result across repeated runs or different task mixes.
Artificial Analysis reports that Terra guessed on 38% of questions it could not answer in its hallucination test, down from 88% for GPT-5.6 Terra. That is a result within that test, not a guarantee that Terra will avoid unsupported answers in general use. OpenAI says Terra meets its Preparedness Framework’s Critical threshold in cybersecurity and retains safeguards developed for GPT-6 Astra; that classification is OpenAI’s, not an independent assessment in the supplied material.
Testing Costs on Actual Workloads
The next useful comparison for organizations is to run Terra alongside their current model on representative tasks, recording both successful completions and the cost of retries, review and corrections. Teams considering adoption will also need to decide which effort setting suits each job; the benchmark curve indicates that paying for maximum effort can raise costs substantially without matching the performance of leading larger models.
OpenAI has made the model available through its API, ChatGPT Work and Codex, so users can begin evaluating it in those environments. Further independent measurements—particularly for speed and performance across real-world workflows—could clarify whether Terra’s benchmark cost advantage holds beyond the tests available at launch. Until then, the evidence supports a qualified conclusion: Terra may be a practical option for routine, high-volume tasks, while harder work may still call for a more capable model.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a model OpenAI released on October 26, 2026. The company positions it for communication and computer-based workplace tasks, between Luna and its larger Sol and Astra models.
Is Terra OpenAI’s highest-scoring model?
No. Artificial Analysis scored Terra at 48.1 on its Intelligence Index at maximum effort, below several models in its comparison, including GPT-6.1 Sol and Claude Opus 5.5.
Why might Terra be cheaper than a model with lower token prices?
The relevant comparison can be cost per completed task, not just price per token. A model’s total cost depends on how many input, output and reasoning tokens it uses. Artificial Analysis reported that Terra at low effort cost less per task than Luna in its evaluation.
Which Terra effort setting is cheapest?
In the reported Artificial Analysis figures, low effort cost $0.055 per task. Non-reasoning cost $0.107, nearly twice as much, while scoring lower. Costs and scores reflect that benchmark’s tasks and may differ for other workloads.
Where can users access GPT-6.1 Terra?
OpenAI says Terra is available through the API, ChatGPT Work and Codex. The supplied launch information lists access for Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex.
Source: ThorstenMeyerAI.com
