🔍 Read the full analysis: GPT-6.1 Terra: OpenAI’s Potential Budget Pick For Getting Work Done 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, positioning it as a lower-cost model for routine work in ChatGPT Work, Codex and the API. Artificial Analysis measured a $0.40 average task cost and a 48.1 Intelligence Index score at max effort; its results also show low and medium effort offer better cost efficiency than higher settings. Terra does not lead the benchmark, and the source’s comparisons come from different evaluation sets.
OpenAI released GPT-6.1 Terra on Oct. 26, a mid-tier model it says is designed for everyday communication and computer-based work, with API prices of $1 per million input tokens and $5 per million output tokens. Independent benchmark firm Artificial Analysis measured Terra at $0.40 per task and 48.1 on its Intelligence Index at maximum effort, making it a lower-cost option than larger models in that comparison, but not the highest-scoring one.
Terra is available to eligible 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, according to the source report. OpenAI lists cached input at $0.05 per million tokens and says Terra’s API prices are 50% lower than GPT-5.6 Terra’s. It describes the model as suited to high-volume tasks such as reading invoices and entering their details into accounting software.
Artificial Analysis’s launch-day evaluation put Terra’s max-effort score at 48.1 on its Intelligence Index, which combines ten evaluations. The firm reported an average cost of $0.40 per task at that setting. 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 figures describe performance and average task costs on Artificial Analysis’s index, not a guaranteed price for every user’s workload.
The source report says Artificial Analysis measured six effort settings, with scores from 33.5 in non-reasoning mode to 48.1 at max effort. It reports that low effort cost $0.055 per task and scored 36.2, while maximum effort cost $0.40. The report characterizes low and medium settings as the strongest points on Terra’s cost-efficiency curve. It also notes that non-reasoning mode was not the cheapest: its reported $0.107 average task cost was higher than low effort’s, despite a lower score.
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 Comparison
Terra’s launch puts attention on a metric that can matter more to businesses than a model’s listed token rate: the cost of finishing a task. A cheaper rate per token does not necessarily mean a cheaper result if a model uses more tokens or needs additional attempts. Artificial Analysis’s reported figures suggest Terra can be economical for certain workloads when used at lower effort settings, although the result depends on the task mix represented by its benchmark.
For teams processing many routine requests, lower average costs could affect which model they assign to each job and how far an API budget goes. The trade-off is performance: Artificial Analysis’s max-effort score for Terra remains below several larger models in its comparison, and the source report says OpenAI recommends Sol or Astra for the most complex coding tasks. The evidence supports testing Terra against a company’s own work, rather than assuming one model is the best choice for every task.
Individual users may also see more tasks covered by an allowance. OpenAI says ChatGPT Work and Codex users can complete substantially more tasks with Terra than with Sol or Astra under the same allowance. That is a company statement; the report does not provide a universal task count, and actual usage will vary with task length, settings and product limits.
Popular right now
LG 85-inch QNED75B Mini LED Smart TV Bundle with HDMI and CPS Protection Plan
Sony WH-1000XM4C Wireless Noise-Cancelling Headphones, Black
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, KSM150PS, Almond Cream
As an affiliate, we earn on qualifying purchases.
Where Terra Sits in OpenAI’s Lineup
OpenAI positions Terra between GPT-6 Luna, the smaller model, and the larger Sol and Astra models. The launch is an update to the Terra name rather than a claim that this model leads the field. In Artificial Analysis’s reported table, Luna scored 38.1 at $0.07 per task, while Terra reached 48.1 at $0.40 and Sol scored 51.8 at $0.72. Those results show a performance-cost gradient, not a single winner across all use cases.
The source report also cites OpenAI’s own evaluations, which should be distinguished from the independent benchmark. OpenAI said Terra at medium effort scored 57.6% on Agents’ Last Exam, compared with 42.6% for GPT-5.6 Terra, at reported task costs of $0.29 and $0.84 respectively. On Terminal-Bench 4.0, OpenAI reported a 50.3% score for Terra at max effort, compared with a previously published 21.5% for GPT-5.6 Terra. These are company-reported test results, not Artificial Analysis findings.
OpenAI also reported lower error rates on its internal error-inducing prompt set and said Terra completes tasks more than three times faster than Astra and Sol. The source notes that Artificial Analysis had not published speed measurements for Terra at the time its report was checked. OpenAI further said Terra meets the “Critical” threshold in cybersecurity under its Preparedness Framework; that classification is OpenAI’s own assessment.
Limits of the Benchmark Evidence
The reported benchmark figures do not show how Terra will perform across every company’s workload. Artificial Analysis’s cost is an average across its evaluation tasks, while real costs can vary with prompt size, output length, cached input, reasoning effort and the number of calls needed to complete a job. The source report does not include the full results for all six settings in the material provided, so comparisons beyond the listed non-reasoning, low and max figures should not be inferred.
Several favorable results cited in the report come from OpenAI’s own tests, including the speed, coding and internal error-rate comparisons. The source also says independent speed measurements were not available at the time of its check. OpenAI’s cybersecurity classification is likewise the company’s assessment, not an independently verified finding presented in the report.
It remains unclear how broadly Terra’s lower-cost profile will translate to production systems, how its performance changes across specific industries, and what product-level usage limits apply to different ChatGPT plans. The report also identifies Xiaomi’s open-weight MiMo-V2.6-Pro as scoring 46.3 for $0.13 per task in Artificial Analysis’s table, but those benchmark results alone do not establish equivalent reliability, deployment costs or suitability for a particular buyer.
What Teams Should Test Next
The immediate next step for companies considering Terra is to compare it with their current model on representative tasks, tracking cost per successfully completed task alongside accuracy, latency and the need for human review. The source report recommends starting at low or medium effort and routing more difficult work to larger models where needed. That is a practical approach suggested by the reported benchmark curve, not a guarantee of savings.
OpenAI has made Terra available through its listed ChatGPT Work, Codex and API channels, so teams can begin evaluating it now. Buyers should verify current pricing, plan allowances and model access directly with OpenAI, since those product terms can change. Further independent speed measurements or broader evaluations could clarify whether the company’s efficiency claims hold across different workloads.
For users, the main near-term question is whether Terra’s lower-cost positioning results in more completed work within their existing product allowance without sacrificing acceptable quality. OpenAI says it will do so, but the source material provides no universal task quota or independent measurement of allowance usage. The next useful evidence will come from real-world use and additional benchmark data, particularly for complex tasks where the source report says larger models remain preferable.
Key Questions
What is GPT-6.1 Terra?
It is a mid-tier OpenAI model released on Oct. 26, 2026, for communication and computer-based work. OpenAI lists it in ChatGPT Work, Codex and the API.
How much does GPT-6.1 Terra cost?
OpenAI’s 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 reported an average task cost of $0.40 at max effort on its benchmark; that is a benchmark average, not a fixed price for every task.
Does Terra outperform OpenAI’s larger models?
Not across the reported comparisons. Artificial Analysis scored Terra at 48.1 at max effort, below GPT-6.1 Sol at 51.8 and GPT-6 Astra at 52.7. OpenAI says Sol and Astra remain better choices for its most complex coding tasks.
Which effort setting appears most cost-efficient?
The source report says Artificial Analysis found low and medium effort to be on Terra’s cost-efficiency frontier. Its listed low-effort result was $0.055 per task and a score of 36.2; higher settings cost more, with max effort reaching $0.40 and 48.1. The best setting for a specific user depends on the task.
Are the performance and speed claims independently verified?
Some are, and some are not. Artificial Analysis published an independent evaluation, while results such as OpenAI’s speed comparison and internal error-rate tests are company-reported. The source report said Artificial Analysis had not published speed measurements for Terra at the time it checked.
Source: ThorstenMeyerAI.com
