🔍 Read the full analysis: How GPT-6.1 Terra Compares When Cost Matters More Than Peak Intelligence on ThorstenMeyerAI.com
Get audio and creator gear 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 its API at $1 per million input tokens and $5 per million output tokens. Artificial Analysis measured a $0.40 average task cost and a 48.1 Intelligence Index score at maximum effort; its results suggest low and medium effort offer better cost-efficiency, while stronger models score higher on the index.
OpenAI released GPT-6.1 Terra on October 26, pitching it as a lower-cost model for everyday communication and computer-based work. Its API price is $1 per million input tokens and $5 per million output tokens, while independent benchmarker Artificial Analysis measured an average cost of $0.40 per task at maximum effort and a 48.1 Intelligence Index score—below several more capable models.
OpenAI says Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, as well as through the API under the name gpt-6.1-terra. The listed API rates are half the rates given for its predecessor, GPT-5.6 Terra: $1 versus $2 per million input tokens, and $5 versus $12 per million output tokens. Cached input is priced at $0.05 per million tokens. OpenAI says the model is intended for routine work such as reading invoices and entering information into accounting software.
Artificial Analysis’s launch-day evaluation places Terra at 48.1 points and $0.40 per task at maximum effort. In the same index, GPT-6.1 Sol scored 51.8 at $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. The index combines ten evaluations and reports an average task cost across them. These results describe that benchmark mix; they do not establish what a particular customer will pay for a specific workload.
Terra has six effort settings, and the measured cost and score change with the setting. Artificial Analysis reports a range of 33.5 to 48.1 points and $0.055 to $0.40 per task, from low to maximum effort. Its summary says low and medium settings sit on its cost-efficiency frontier, meaning no tested model in that comparison was both cheaper and higher-scoring. The provided source does not include every setting’s full score and cost data, so figures beyond those specifically reported should not be inferred.
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
The Cost of Finishing a Task
For organizations buying AI at scale, the headline price per token can obscure the cost of getting usable work completed. A model may use fewer tokens to finish a task, or may need more reasoning tokens at a higher effort setting. Artificial Analysis’s reported comparison therefore shifts attention from unit price to average task cost, although its benchmark average is not a universal estimate for businesses.
The trade-off is capability. Terra’s maximum-effort score of 48.1 is below Sol’s 51.8 and several other models in the published index. A lower cost may suit repeatable, lower-risk work, but the results do not show that Terra can replace higher-scoring models for every task. Companies may find it useful to test the model on their own workload and compare both quality and total cost before changing deployment choices.
That distinction also matters to individual users: cheaper tokens do not automatically mean that every prompt is cheaper or that the answer will be sufficient. The relevant comparison depends on task complexity, output length, effort level and the cost of correcting errors.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Where Terra Fits in OpenAI’s Lineup
OpenAI positions Terra between its smaller GPT-6 Luna and larger Sol and Astra models. Artificial Analysis scored Luna at 38.1 and listed an average cost of $0.07 per task, compared with Terra’s 48.1 and $0.40 at maximum effort. Those numbers do not support a simple claim that Terra is cheaper for every task. The source’s author argues that Terra at low effort can cost less per completed task than Luna because it produces fewer tokens, but the supplied material does not give the full underlying task-level comparison.
OpenAI’s own launch-page charts provide separate results from Artificial Analysis’s independent index. OpenAI reports that Terra at medium effort scored 57.6% on Agents’ Last Exam for $0.29 per task, compared with GPT-5.6 Terra’s 42.6% and $0.84. It also reports a 50.3% score on Terminal-Bench 4.0 at maximum effort, compared with 21.5% for GPT-5.6 Terra. These are OpenAI-reported measurements, not independent evaluations, and their task-specific scores should not be conflated with the Intelligence Index.
OpenAI also says Terra completes tasks more than three times faster than Sol and Astra, and that its error rate fell on the company’s internal set of error-inducing prompts. Artificial Analysis had not published speed measurements for Terra when the source checked. OpenAI’s claims about speed and internal error testing have not been independently established in the supplied material.
Benchmark Results Have Limits
The benchmark results do not establish how Terra will perform across all users’ applications. Artificial Analysis’s score is an aggregate across ten evaluations, and its average task cost reflects that test mix. Actual spending can vary with prompts, output length, caching, reasoning effort and how often a task must be retried or corrected.
Some claims in the launch material remain company-reported. The supplied source gives no independent verification of OpenAI’s speed comparison, its internal factual-error results, or its cybersecurity classification under the Preparedness Framework. It also does not provide complete figures for every effort setting, including the full set of scores, costs and hallucination rates. The reported reduction in guessed answers on Artificial Analysis’s hallucination test—from 88% for GPT-5.6 Terra to 38% for Terra—does not mean the new model never makes unsupported claims.
It is also unclear from the available material how Terra’s pricing and access will change over time, or how its average cost will compare on specific customer workflows. The benchmark figures are dated October 26, 2026, and should be read as a snapshot rather than a guarantee of future price or performance.
What Buyers Should Test Next
Businesses considering Terra can compare it with their current models using representative tasks, recording cost per completed task, accuracy, latency and the amount of human review required. Testing low and medium effort first may help establish whether the model meets a task’s quality threshold before teams pay for higher effort or route work to Sol or Astra.
Readers should also watch for additional independent benchmarks, particularly speed measurements and fuller results across Terra’s six effort settings. OpenAI’s launch makes the model available through its stated ChatGPT plans and API, but the supplied announcement does not identify a future release milestone or scheduled benchmark update.
Key Questions
What does GPT-6.1 Terra cost?
OpenAI lists API prices of $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Actual cost depends on usage and the task.
Is Terra OpenAI’s highest-scoring model?
No. In Artificial Analysis’s October 26, 2026, Intelligence Index, Terra scored 48.1 at maximum effort, below GPT-6.1 Sol at 51.8 and several other listed models.
Which Terra effort setting is the best value?
Artificial Analysis said its low and medium settings sit on the cost-efficiency frontier in the tested comparison. The best setting for a specific user depends on whether its quality is sufficient for the task.
Does the benchmark show Terra is cheaper than Luna?
Not across all settings in the supplied figures. Artificial Analysis lists Luna at $0.07 per task and Terra at maximum effort at $0.40. The source says Terra at low effort can cost less per task because it uses fewer tokens, but the complete comparison data was not supplied.
Where can users access GPT-6.1 Terra?
OpenAI says Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and through the API as gpt-6.1-terra.
Source: ThorstenMeyerAI.com
