🔍 Read the full analysis: GPT-6.1 Terra Isn’t OpenAI’s Smartest Model, But Could It Be The Best Value? on ThorstenMeyerAI.com
Get movie nights 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 its API at $1 per million input tokens and $5 per million output tokens. Artificial Analysis measured a 48.1 Intelligence Index score and an average task cost of $0.40 at maximum effort; its results suggest low and medium settings offer the strongest cost-efficiency, not the highest score.
OpenAI released GPT-6.1 Terra on Oct. 26, a model priced below its larger GPT-6.1 models and aimed at routine communication and computer-based work. In an evaluation published the same day, Artificial Analysis measured Terra at 48.1 on its Intelligence Index and an average of $0.40 per task at maximum effort, below several higher-scoring models in cost but also in performance.
OpenAI lists Terra’s API price at $1 per million input tokens and $5 per million output tokens, compared with $2 and $12 for GPT-5.6 Terra, according to the launch report. Cached input costs $0.05 per million tokens. OpenAI says the model is available in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, as well as through the API under the name gpt-6.1-terra.
Artificial Analysis’s index combines 10 evaluations and tracks the cost of completing them, including reasoning and cached input tokens. Its reported figures put GPT-6.1 Terra at maximum effort at 48.1 points and $0.40 per task. 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 index’s evaluation tasks, not a guaranteed price or result for every customer workload.
The report says Terra has six effort settings, ranging from non-reasoning to maximum effort. The supplied figures show the low setting at 36.2 points and $0.055 per task, while maximum reaches 48.1 points for $0.40. The report characterizes low and medium as points on the cost-efficiency frontier: among tested models and settings, no option is both cheaper and higher-scoring. The provided data also say the non-reasoning setting costs nearly twice as much per task as low effort while scoring lower, a reminder that token prices alone do not determine total task cost.
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
Why Task Costs Change the Calculation
Terra’s launch puts attention on a practical question for companies adopting AI: how much does a completed task cost? A model with a lower price per token can still use more tokens to finish work. Artificial Analysis’s measurements indicate that, for its evaluated tasks, Terra at low effort can cost less per task than GPT-6 Luna despite Luna’s lower token price. That result depends on the benchmark workload and should not be treated as a universal cost comparison.
For businesses processing many similar requests, the difference between a low- and maximum-effort setting could shape operating budgets. The report’s figures show maximum effort costing about 7.4 times as much as low effort for 12 additional Intelligence Index points. That may support routing routine jobs to a less expensive setting and reserving larger models or higher effort for difficult work, but the right choice will depend on accuracy requirements, task mix and actual usage.
Terra is not the highest-scoring model in the cited comparison. Artificial Analysis lists Claude Opus 5.5, Kimi K3.1 Pro and other models above it. The report also identifies Xiaomi’s open-weight MiMo-V2.6-Pro as scoring 46.3 for $0.13 per task, ahead of Terra’s high and extra-high settings on both score and cost. The release is therefore best understood as a pricing and workload option, not evidence that Terra leads the field.
Popular right now
Ninja Flip Toaster Oven and Air Fryer, 8-in-1, Stainless Steel, SP151
KitchenAid Classic Series 4.5 Quart Tilt-Head Stand Mixer K45SS, Onyx Black
Roborock Robot Vacuum and Mop with LiDAR Navigation and 10,000 Pa Suction
As an affiliate, we earn on qualifying purchases.
Terra Between Luna and Sol
OpenAI describes Terra as a “fast, efficient model” for communication and computer-based work, with invoice reading and entry into accounting software as an example of high-volume use. The report places it between the smaller GPT-6 Luna and the larger Sol and Astra models. OpenAI says the same ChatGPT Work allowance can cover substantially more Terra tasks than Sol or Astra, though the source does not provide a quantified allowance comparison.
OpenAI’s own launch-page charts, as summarized by ThorstenMeyerAI.com, report improvements over GPT-5.6 Terra on selected tasks. On Agents’ Last Exam, Terra at medium effort scored 57.6% for $0.29 per task, compared with 42.6% and $0.84 for GPT-5.6 Terra at medium effort, according to the report. At maximum effort, OpenAI’s chart lists Terra at 60.0% and Claude Opus 5.5 at 63.2% on that evaluation, with respective task costs of $1.01 and $8.82.
The same report cites OpenAI figures showing Terra at maximum effort solving 50.3% of Terminal-Bench 4.0 tasks, compared with a published 21.5% for GPT-5.6 Terra. OpenAI also says Terra completes tasks more than three times faster than Astra and Sol. Those performance and speed statements come from OpenAI; Artificial Analysis had not published speed measurements for Terra at the time covered by the report.
Limits of the Published Comparisons
The available figures do not establish how Terra will perform across every customer’s workload. Artificial Analysis’s cost per task is an average across its evaluation tasks; actual spending can vary with prompt length, output length, caching, reasoning settings and the frequency of retries. The supplied material does not include enough detail to independently reproduce the benchmark results or assess how representative the tests are of routine business use.
Some of the effort-setting table is missing from the source text, including the medium, high and extra-high rows. The report summarizes the overall curve and gives low and maximum figures, but those omitted entries prevent a full check of each setting’s score, cost and hallucination rate here. OpenAI’s claims about speed, factual errors and cybersecurity classification are also company-reported, not measurements confirmed by Artificial Analysis in the material provided.
The source reports that Terra guessed on 38% of questions it could not answer in Artificial Analysis’s hallucination test, compared with 88% for GPT-5.6 Terra. The test’s exact sample size and methods are not included in the supplied material. The result should be read as a benchmark finding, not a guarantee that Terra will avoid unsupported answers in other settings.
How Teams May Test Terra
Terra is available through OpenAI’s API and in ChatGPT Work and Codex for the listed paid plans. The next useful evidence will come from independent speed measurements and broader real-world use, as well as comparisons on specific workloads such as invoice processing, coding and office-agent tasks. The source does not give a schedule for further benchmark releases.
Organizations considering Terra can compare it with existing models on representative tasks, measuring not only token charges but also completion rates, review time, error costs and total spend per successful result. The launch report’s suggested starting point is low or medium effort for routine work, with more capable models or higher effort reserved for harder tasks. That remains a practical recommendation to test, rather than a result established for every organization.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a newly released OpenAI model intended for communication and computer-based work. OpenAI offers it in ChatGPT Work and Codex for specified paid plans and through its API.
How much does GPT-6.1 Terra cost?
The reported 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 an average task cost of $0.40 at maximum effort across its index evaluations; that figure is not a fixed price for every task.
Is Terra OpenAI’s highest-scoring model?
No. In the reported Artificial Analysis comparison, Terra scored 48.1 at maximum effort, below GPT-6.1 Sol at 51.8 and several other listed models. OpenAI’s launch material also says Sol and Astra remain better choices for its most complex coding tasks.
Which Terra effort setting offers the best value?
Artificial Analysis’s report identifies low and medium effort as being on its cost-efficiency frontier. The supplied source gives low effort as 36.2 points for $0.055 per task and maximum effort as 48.1 points for $0.40; which setting is best depends on the work and the required level of performance.
Can Terra replace larger models for every task?
The available evidence does not support that conclusion. Terra costs less in the cited comparison, but larger models score higher, and OpenAI says Sol and Astra are preferable for its most complex coding work. Teams would need to test Terra against their own tasks and quality requirements.
Source: ThorstenMeyerAI.com
