🔍 Read the full analysis: GPT-6.1 Terra: Affordable AI For Tasks That Don’t Need The Smartest Model on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- 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 as a lower-cost model for routine work rather than its strongest option. Artificial Analysis measured a $0.055 average task cost at low effort and a score of 36.2; at maximum effort, Terra scored 48.1 for $0.40 per task, below several more capable models.
OpenAI released GPT-6.1 Terra on October 26, a model aimed at routine communication and computer-based work, with API prices half those listed for its predecessor, GPT-5.6 Terra. Independent benchmark firm Artificial Analysis found that Terra’s low and medium effort settings offer a strong cost-to-score trade-off on its evaluation, while the model’s maximum-effort score remains below several more capable systems.
OpenAI lists Terra at $1 per million input tokens and $5 per million output tokens, compared with $2 and $12 for GPT-5.6 Terra. Cached input is priced at $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. The company describes it as a fast, efficient option for work such as processing incoming invoices and entering information into accounting software.
Artificial Analysis’s launch-day evaluation gives Terra a 48.1 Intelligence Index score at maximum effort, with an average cost of $0.40 per task across its benchmark evaluations. For comparison, the firm lists GPT-6.1 Sol at 51.8 and $0.72 per task, and Claude Opus 5.5 at 57.6 and $5.98. Those figures reflect the index’s measured tasks and pricing assumptions, not a guarantee of the cost or quality of every real-world job.
The six-setting results show a trade-off between reasoning effort and cost. Artificial Analysis measured Terra at 36.2 points and $0.055 per task at low effort, rising to 48.1 points and $0.40 at maximum effort. Its non-reasoning setting scored 33.5 and cost $0.107 per task, so it was not the cheapest option in these tests. The source says low and medium settings sit on the benchmark’s cost-efficiency frontier; higher settings cost more without matching that efficiency.
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 Model Choice
Terra’s launch highlights why a per-token price alone can mislead buyers comparing AI services. A model that writes fewer tokens can cost less to complete a task even when another model has a lower listed price per token. In Artificial Analysis’s measurements, GPT-6 Luna costs less per token, but Terra at low effort had a lower reported average cost per task. That result applies to the benchmark’s task mix and should not be assumed for every workload.
For businesses handling many routine requests, the relevant calculation is the cost of a completed task at an acceptable quality level. Terra’s lower settings could make sense for structured, repeatable work, while more demanding coding or reasoning jobs may justify routing requests to a stronger model. OpenAI itself says Sol and Astra remain better choices for its most complex coding tasks. The practical question for adopters is whether savings hold on their own data, including the cost of errors, human review and retries.
The benchmark also shows that Terra is not the top-scoring model in the comparison. Artificial Analysis lists several higher-scoring systems, while its results also include lower-cost alternatives, including open-weight MiMo-V2.6-Pro. Terra’s selling point in these figures is a balance of capability and task cost, not leadership on raw intelligence scores.
Products Worth Considering
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Terra Fits OpenAI’s Lineup
OpenAI positions Terra between its smaller Luna model and larger Sol and Astra models. The launch information says Terra’s API rates are 50% lower than GPT-5.6 Terra’s and GPT-6.1 Sol’s. OpenAI also says eligible ChatGPT Work and Codex users can complete substantially more tasks within the same allowance than with Sol or Astra; that is a company claim, and the material does not provide a comparable independent allowance test.
OpenAI’s own launch-page measurements report improvements on selected benchmarks. At medium effort, Terra scored 57.6% on Agents’ Last Exam, compared with 42.6% for GPT-5.6 Terra, at a reported $0.29 per task versus $0.84. On Terminal-Bench 4.0 at maximum effort, OpenAI reports a 50.3% task score for Terra, against 21.5% for its predecessor’s published result. The company also reports lower error rates on its internal error-inducing prompt set. These are OpenAI-reported results, not the same evaluation as Artificial Analysis’s independent index.
Artificial Analysis says its Intelligence Index combines ten evaluations, including agentic office work, terminal tasks, science and hallucination testing, and records associated task costs. Its figures cited here were retrieved on October 26, 2026. The firm’s scores and average costs provide a comparative snapshot, but performance can vary by task, prompt, configuration and price changes.
Real-World Costs Still Vary
The supplied measurements do not establish how Terra will perform across every company’s workload. Artificial Analysis’s average task cost depends on its evaluation set, while OpenAI’s figures come from company-selected benchmarks and, in one case, an internal prompt set. Independent speed measurements were not available in the source material, despite OpenAI’s claim that Terra completes tasks more than three times faster than Astra and Sol.
It is also unclear how much users will save in ChatGPT Work or Codex in practice, since the source does not specify the allowance calculation or provide results across different usage patterns. The launch material says Terra meets OpenAI’s Critical threshold in cybersecurity under its Preparedness Framework; that is OpenAI’s own classification, not an external assessment. The benchmark comparisons do not establish how often Terra’s answers will require correction in production.
Testing Terra on Routine Work
The next useful evidence will come from independent evaluations of Terra on additional tasks, including speed, reliability and performance under real workloads. Businesses considering adoption can compare cost per completed task across Terra’s effort settings and stronger alternatives using their own prompts, while tracking accuracy, review time and retries alongside API charges.
OpenAI’s launch makes Terra available through the listed ChatGPT Work, Codex and API channels, so users can begin testing it against existing workflows. The available source gives no schedule for further model updates or new benchmark results. Until those emerge, the measured case is strongest for routine tasks where low or medium effort is sufficient; demanding work may still call for a larger model.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a model OpenAI released on October 26, 2026, for communication and computer-based tasks. It is positioned between the smaller Luna model and larger Sol and Astra models.
How much does GPT-6.1 Terra cost?
OpenAI lists API pricing at $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 usage and the effort setting.
Is Terra OpenAI’s most capable model?
No. Artificial Analysis scored Terra at 48.1 at maximum effort, below GPT-6.1 Sol at 51.8 and several other models in its comparison. OpenAI says Sol and Astra are better choices for its most complex coding tasks.
Which Terra setting was cheapest in the independent test?
Low effort had the lowest reported average cost, $0.055 per task, with an Intelligence Index score of 36.2. The non-reasoning setting cost $0.107 per task in the same evaluation, so the lowest reasoning setting was not the least expensive.
Are the benchmark results a guarantee of savings?
No. The reported average costs reflect Artificial Analysis’s evaluation tasks and stated price assumptions. Results in a particular business workflow may differ based on the task, output length, accuracy needs and amount of human review.
Source: ThorstenMeyerAI.com
