🔍 Read the full analysis: What Makes GPT-6.1 Terra A Potentially Affordable Workhorse? 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 everyday workplace and computer-based tasks. Artificial Analysis found that Terra’s low and medium reasoning settings offered the strongest cost-to-score trade-off in its evaluation, but higher-scoring models remain available for harder work, and some cheaper alternatives beat Terra on both cost and score.
OpenAI released GPT-6.1 Terra on Oct. 26, presenting it as a lower-cost model for workplace and computer-based tasks. Its API price is $1 per million input tokens and $5 per million output tokens, half the listed rates for its predecessor, GPT-5.6 Terra; independent testing also found that Terra’s low and medium reasoning settings delivered a strong cost-to-score balance on the measured tasks.
Terra is positioned between OpenAI’s smaller Luna model and its larger Sol and Astra models. OpenAI says it is available to 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. The company describes it as suited to routine communication and computer work, citing invoice reading and data entry as examples. OpenAI says its subscription allowance can cover substantially more tasks with Terra than with Sol or Astra.
Artificial Analysis, a benchmarking firm, scored Terra at 48.1 on its Intelligence Index at maximum effort, at an average cost of $0.40 per task across the index’s evaluation tasks. The firm’s figures put GPT-6.1 Sol at 51.8 and $0.72 per task, and Claude Opus 5.5 at 57.6 and $5.98. These are benchmark averages, not guaranteed prices or results for every user’s workload. At maximum effort, Terra scores below those models, despite costing less per evaluated task.
The evaluation’s six effort settings show why a single token-price figure can mislead: cost depends on how many tokens a model uses to complete a task. Artificial Analysis reports Terra at low effort scoring 36.2 for an average of $0.055 per task, while its non-reasoning setting costs $0.107 and scores 33.5. The source says low and medium sit on the evaluation’s cost-efficiency frontier; settings above them cost more without improving the measured trade-off against other tested options. The supplied figures do not include the complete six-setting table.
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 Cost Changes the Comparison
For organizations running AI repeatedly across office processes, cost per completed task can matter more than the posted price per token. A model with a higher token rate may still be cheaper for a particular job if it uses fewer tokens; conversely, a low rate does not guarantee a low total bill. Artificial Analysis says Terra’s low-effort setting costs less per task than Luna in its evaluation, despite Luna’s much lower listed token price. That finding applies to the benchmark’s tasks, not automatically to every workflow.
The model also creates a practical choice between expense and capability. Terra’s benchmark score rises with effort, but the source reports that its maximum setting costs 7.4 times as much as low effort for 12 additional index points. Businesses may be able to reserve more capable, expensive models for complex cases and use Terra for routine work. That routing strategy could control spending, but its success depends on a company’s own task mix, accuracy requirements and integration costs.
Lower cost does not make Terra the strongest model in the supplied results. Artificial Analysis lists higher scores for Claude Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon. It also reports that Xiaomi’s open-weight MiMo-V2.6-Pro scored 46.3 for $0.13 per task, outperforming Terra’s high and extra-high settings on both measures. The launch is best understood as another option in a crowded market, not proof that Terra leads it.
Popular right now
TCL 85 Designer Series QD-Mini LED TV (Model 85A400U) with Google Smart TV
Dyson Cyclone V10 Origin Cordless Vacuum
Philips 4400 Series Fully Automatic Espresso Machine with LatteGo Milk System
As an affiliate, we earn on qualifying purchases.
Terra’s Place in OpenAI’s Lineup
OpenAI’s launch presents Terra as a workhorse model, rather than its top-performing option. Its posted API rates are 50% lower than GPT-5.6 Terra’s: the earlier model is listed at $2 per million input tokens and $12 per million output tokens, compared with Terra’s $1 and $5. Cached input for the new model is listed at $0.05 per million tokens. Artificial Analysis separately reports its own average cost-per-task estimates, which should not be confused with those token prices.
OpenAI also published company-run results for particular tests. On Agents’ Last Exam, it says Terra at medium effort scored 57.6% and cost $0.29 per task, compared with GPT-5.6 Terra’s 42.6% and $0.84. For Terminal-Bench 4.0, OpenAI reports that Terra at maximum effort solved 50.3% of tasks, versus 21.5% for GPT-5.6 Terra. Those figures come from OpenAI’s own launch materials and use specific benchmarks; they are distinct from Artificial Analysis’s broader independent index.
Other company claims include faster task completion than Sol and Astra, and fewer errors on OpenAI’s internal prompt set. OpenAI also says Terra meets the “Critical” threshold in cybersecurity under its Preparedness Framework. These are the company’s statements, not independent confirmations in the supplied material. Artificial Analysis reports a separate reduction in guessing on questions Terra could not answer: 38% for the new model, compared with 88% for GPT-5.6 Terra in its hallucination test.
Limits of the Launch-Day Evidence
Benchmark results do not establish performance on every real-world task. Artificial Analysis’s cost figures are averages across its evaluation tasks, while OpenAI’s results come from company-selected tests. The materials provided do not show how Terra performs across different industries, languages, software setups or long-running workflows, nor do they establish whether reported savings hold once implementation and review costs are included.
Some comparisons remain incomplete. Artificial Analysis had not published speed measurements for Terra when the source checked on Oct. 26, 2026, so OpenAI’s claim that Terra is more than three times faster than Sol and Astra is not independently corroborated there. The source also supplies only part of the six-setting effort table, preventing a full verification of every stated score, cost and error-rate point. It is not clear how widely the launch availability is rolling out by region or whether access limits differ among plans.
How Buyers Can Test Terra
The immediate next step for potential users is to compare Terra with alternatives on their own representative tasks, recording both completion quality and total cost. For businesses, that means testing invoice processing, coding or other intended workflows at low and medium effort before deciding whether higher settings or a larger model are warranted. Teams should also check current API prices and product availability directly, as launch-day details may change.
Further evidence may come from independent speed measurements, additional benchmark coverage and customer reports as Terra is used in production. Until those results are available, the clearest supported conclusion is limited: launch-day testing indicates that Terra can be a cost-effective option at some effort levels, while harder tasks may still call for more capable models.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model aimed at workplace communication and computer-based tasks. OpenAI offers it through ChatGPT Work, Codex and its API, according to the launch report.
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. Actual spending depends on token use; Artificial Analysis’s average task costs are separate benchmark estimates.
Is Terra OpenAI’s most capable model?
No such claim is supported by the supplied results. Artificial Analysis scores Terra below GPT-6.1 Sol and several competing models at the settings compared, and OpenAI says Sol and Astra remain better for its most complex coding tasks.
Which Terra setting appears most cost-efficient?
Artificial Analysis places Terra’s low and medium effort settings on its cost-efficiency frontier for the evaluated models and tasks. That result is benchmark-specific; users should test settings against their own accuracy and workload needs.
What remains unverified?
The supplied material does not include independent speed measurements for Terra or a complete table for all six effort settings. OpenAI’s speed, internal error-rate and cybersecurity statements are company claims, and performance across real customer workflows remains to be established.
Source: ThorstenMeyerAI.com
