🔍 Read the full analysis: Is GPT-6.1 Terra The Right OpenAI Model For Everyday Jobs? 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 October 26, 2026, pricing API use at $1 per million input tokens and $5 per million output tokens. Artificial Analysis found its lower and medium effort settings offer a favorable balance of task cost and benchmark score, while higher effort or more complex work may call for a stronger model.
OpenAI released GPT-6.1 Terra on October 26, describing it as a fast, efficient model for everyday communication and computer-based work. Its API price is $1 per million input tokens and $5 per million output tokens, half the listed price of GPT-5.6 Terra, while independent testing suggests that the cost of completing a task—not the token rate alone—is the more useful measure for comparing models.
Terra is positioned between OpenAI’s smaller GPT-6 Luna 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. OpenAI says the same ChatGPT allowance can cover substantially more tasks with Terra than with Sol or Astra; that is the company’s description, not an independently verified allowance comparison.
Artificial Analysis, an independent benchmarking firm, published results on launch day. It scored Terra at 48.1 on its Intelligence Index at maximum effort, below GPT-6.1 Sol’s 51.8 and Claude Opus 5.5’s 57.6. The firm put Terra’s average cost across its evaluation tasks at $0.40, compared with $0.72 for Sol and $5.98 for Opus 5.5. These figures reflect the benchmark’s tasks and cost methodology, rather than a guaranteed price for every user’s work.
Artificial Analysis measured six effort settings, with scores ranging from 33.5 to 48.1 and average task costs from about $0.05 to $0.40. Its results place the low and medium settings on the firm’s cost-efficiency frontier: tested alternatives did not achieve both a lower task cost and a higher score. The maximum setting cost about 7.4 times as much as low effort for 12 additional index points. The non-reasoning setting was not the cheapest in the test; it cost nearly twice as much per task as low effort while scoring lower.
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
Choosing Terra by Task Cost
Terra’s appeal is not that it beats OpenAI’s larger models on every measure. It is that businesses and individual users may be able to handle routine work at a lower overall cost, particularly when the model uses fewer tokens to complete a task. Artificial Analysis found that GPT-6 Luna costs less per token but that Terra at low effort cost less per task in its evaluation, because Terra generated fewer tokens. That makes per-token rates an incomplete guide to the final bill.
For companies processing large volumes of routine requests, small differences in the cost of a completed task can affect operating budgets. OpenAI cites work such as reading invoices and entering details into accounting software. The benchmark results suggest starting with lower effort for simpler jobs and increasing effort only when the task warrants it. That is a practical interpretation of the test results, not a guarantee that the same settings will suit every workflow.
There are limits to the value proposition. Terra did not lead the Intelligence Index, and Artificial Analysis listed cheaper options that scored well: Xiaomi’s open-weight MiMo-V2.6-Pro scored 46.3 at an average task cost of $0.13. Meanwhile, OpenAI says Sol and Astra remain better choices for its most complex coding tasks. Readers comparing models should weigh performance, reliability, privacy, integration and actual workload costs—not price alone.
Products Worth Considering
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Black
Amazon Kindle Scribe Colorsoft 64GB (1st Generation)
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
What OpenAI and Benchmarks Show
OpenAI describes Terra as a model for the communication and computer-based tasks that fill a workday. In its launch-page measurements, the company says Terra at medium effort scored 57.6% on the Agents’ Last Exam computer-use benchmark, compared with 42.6% for GPT-5.6 Terra at medium effort. OpenAI reported costs of $0.29 and $0.84 per task, respectively, and said Terra matched GPT-6 Astra’s medium-effort score at a lower cost. These are OpenAI’s own benchmark figures; they should be distinguished from Artificial Analysis’s separate evaluation.
OpenAI also reported improved results on its internal error-inducing prompts: the low-effort error rate fell from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra, and the maximum-effort rate fell from 22.3% to 4.3%. On Terminal-Bench 4.0, OpenAI said Terra at maximum effort solved 50.3% of tasks, versus 21.5% for GPT-5.6 Terra. The company said Terra completes tasks more than three times faster than Astra and Sol, but Artificial Analysis had not published speed measurements for Terra as of the source report’s October 26 check.
Artificial Analysis’s Intelligence Index combines ten evaluations, including agentic office work, terminal tasks, science and hallucination, and tracks spending on the tests, including cached input and reasoning tokens. Its launch-day figures describe a particular benchmark suite and model configuration. They do not establish how Terra will perform across every real-world application.
Limits of the Terra Comparisons
The reported results do not settle how Terra will perform for a particular user’s documents, software, or coding tasks. Benchmark scores and average task costs depend on evaluation design, prompting, effort setting and token use; actual costs may differ by workload. The source material also does not provide a complete breakdown of the six-setting results for every metric, so comparisons such as hallucination rates cannot be fully assessed here.
OpenAI’s internal error rates, speed statements and computer-use results are company-reported. The independent evaluation offers a separate comparison but does not verify every OpenAI claim. Artificial Analysis had not released Terra speed measurements at the time its launch-day results were checked. OpenAI’s statement that Terra meets the Critical threshold in cybersecurity under its Preparedness Framework is also the company’s own classification, not an independent finding in the supplied material.
It remains unclear how the model’s performance will hold up in sustained, real-world deployments and whether the reported savings will persist once organizations account for setup, review, errors and integration costs. No release information in the source specifies future pricing changes or a date for additional independent testing.
Testing Terra in Real Workflows
Readers and organizations can compare Terra with existing models by running representative tasks at low and medium effort, then recording completion quality, token use and cost per finished task. More demanding work can be tested at higher settings or against Sol and Astra, while keeping the workload and evaluation criteria consistent. The key next evidence will be independent speed results, broader real-world evaluations and updated benchmark pages as testing expands.
OpenAI has made Terra available through the API and in ChatGPT Work and Codex for the plans identified in its announcement. Further information on deployment experience, pricing and performance remains dependent on what OpenAI and independent evaluators publish after launch.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is an OpenAI model released on October 26, 2026, positioned between Luna and the larger Sol and Astra models. OpenAI says it is designed for routine communication and computer-based work.
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. Cached input is listed at $0.05 per million tokens in the source material. The price of an individual task depends on how many tokens it uses.
Is Terra better than GPT-6.1 Sol?
Not on Artificial Analysis’s overall Intelligence Index at maximum effort: Terra scored 48.1 and Sol scored 51.8. Terra’s average benchmark task cost was lower, at $0.40 compared with Sol’s $0.72, so the choice depends on the task and the balance between cost and performance.
Which effort setting should users start with?
Artificial Analysis’s results suggest trying low or medium effort for routine tasks, because those settings sat on its cost-efficiency frontier. Users should test their own workloads; more difficult tasks may benefit from higher effort or a larger model.
Does Terra always cost less than Luna per task?
No universal comparison is established. In Artificial Analysis’s evaluation, Terra at low effort cost less per task than Luna despite Luna’s lower per-token price, because Terra used fewer tokens. Costs can change with the task, prompts and model settings.
Source: ThorstenMeyerAI.com
