🔍 Read the full analysis: Could GPT-6.1 Terra Be Enough For The Work You Need 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 October 26, 2026, positioning it as a lower-cost model for routine computer-based work. Artificial Analysis measured it at $0.40 per task and an Intelligence Index score of 48.1 at maximum effort; lower effort settings cost less, but the model does not match higher-scoring options on the benchmark.
OpenAI released GPT-6.1 Terra on October 26, a model between its smaller Luna and larger Sol and Astra models, aimed at routine work such as processing invoices. Its API costs $1 per million input tokens and $5 per million output tokens, while Artificial Analysis reported an average cost of $0.40 per benchmark task at maximum effort and an Intelligence Index score of 48.1.
OpenAI describes Terra as a fast, efficient model for communication and computer-based work. It 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 prices are half those of GPT-5.6 Terra, according to the launch material; cached input is priced at $0.05 per million tokens.
Artificial Analysis, which published an evaluation on launch day, measured Terra at 48.1 on its Intelligence Index at maximum effort, averaging $0.40 per task across its evaluation set. On 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 per task. Those scores and costs refer to the benchmark’s combined tasks, not a guarantee of performance or cost for any particular customer’s work.
OpenAI’s own launch charts report gains on selected tests. At medium effort, Terra scored 57.6% on Agents’ Last Exam for $0.29 per task, compared with GPT-5.6 Terra’s 42.6% and $0.84, according to the company’s figures. At maximum effort, OpenAI reported a 50.3% score on Terminal-Bench 4.0. The company also says Terra completes tasks more than three times faster than Sol and Astra, but Artificial Analysis had not published speed measurements for Terra when its launch-day evaluation was checked.
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 the Comparison
Terra’s launch highlights why token price alone can mislead when comparing models. A lower-priced model may use more tokens to finish a task, while a more capable model may cost more but require fewer attempts or less supervision. Artificial Analysis’s task-based figures offer one comparison point: Terra at maximum effort averaged $0.40 per evaluation task, against $0.72 for Sol and $5.98 for Opus 5.5.
The benchmark does not show Terra as the highest-scoring option. Its maximum-effort score of 48.1 trails several tested models, including Opus 5.5, Kimi K3.1 Pro and GPT-6 Astra. The practical question for organizations is whether Terra’s lower measured cost is worth the capability trade-off for their own workloads. OpenAI itself says Sol and Astra remain better choices for the most complex coding tasks.
For teams processing large volumes of repetitive work, a lower cost per completed task could affect how many tasks fit within a budget. But the reported averages are benchmark results, not a business case: actual costs can vary with task complexity, prompt design, retries, input volume and the amount of human review needed.
Products Worth Considering
Logitech MX Keys S Wireless Keyboard, Graphite
Sonos Era 100 Wireless Smart Speaker, Black
Amazon Kindle Paperwhite Signature Edition (12th Generation, 2024), 32 GB, Metallic Black
As an affiliate, we earn on qualifying purchases.
A Middle Model With Adjustable Effort
Terra sits between GPT-6 Luna and the larger Sol and Astra models in OpenAI’s lineup. Its main price distinction is the API rate of $1 per million input tokens and $5 per million output tokens, compared with $2 and $12 for GPT-5.6 Terra, based on the source material. OpenAI says eligible ChatGPT Work users can complete substantially more tasks within the same allowance than with Sol or Astra; that is the company’s description, rather than an independently quantified allowance comparison.
Artificial Analysis evaluated six effort settings, from non-reasoning to maximum. Its reported summary says scores ranged from 33.5 at non-reasoning to 48.1 at maximum, while average task costs ranged from about $0.05 to $0.40. Low and medium effort settings sat on the firm’s cost-efficiency frontier; higher settings did not, meaning other tested options offered a better score-cost combination at those points. Moving from low to maximum effort increased cost about 7.4 times for 12 additional index points, according to the source’s figures.
The benchmark also found a counterintuitive result: non-reasoning was not the cheapest setting. It averaged $0.107 per task, almost twice the $0.055 measured at low effort, and scored lower. The explanation given in the source is that Terra produced fewer output tokens at low effort. That finding is specific to this evaluation and should not be treated as a universal cost pattern.
Real-World Costs Still Vary
The published numbers do not establish how Terra will perform across every company’s workload. Artificial Analysis averages costs across its own evaluation tasks; a task involving long documents, repeated tool use or corrections may have a different cost. The source material does not provide independent speed measurements, despite OpenAI’s claim that Terra is more than three times faster than Sol and Astra.
Some safety and reliability claims also come from OpenAI. The company reports lower error rates on its internal error-inducing prompts and says Terra meets the Critical threshold for cybersecurity under its Preparedness Framework. These are company-reported results and classifications; the material does not include an independent evaluation of those claims. The available figures also do not settle how often users will need to check or correct Terra’s work in production.
Test Terra on Real Workloads
Terra is available now through the API, ChatGPT Work and Codex for the listed customer plans. For organizations considering it, the next useful step is to run representative tasks at low and medium effort, record cost per accepted completion, and compare the results with the larger models they already use. That would test whether the benchmark’s lower-cost pattern applies to their own work.
OpenAI’s launch materials point to continued use of Sol and Astra for demanding tasks, rather than presenting Terra as a replacement for every model. Further independent measurements, particularly on speed and real-world task completion, could clarify how much of Terra’s reported cost advantage carries over beyond benchmark testing.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is OpenAI’s mid-range model, positioned between Luna and the larger Sol and Astra models. OpenAI says it is intended for routine communication and computer-based tasks.
How much 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. Artificial Analysis measured an average of $0.40 per task at maximum effort across its benchmark, a separate measure from token pricing.
Does Terra outperform OpenAI’s larger models?
Not on every reported measure. Artificial Analysis gave Terra a 48.1 Intelligence Index score at maximum effort, below Sol’s 51.8 and Astra’s 52.7 in the cited comparison. OpenAI says Sol and Astra remain better choices for its most complex coding tasks.
Which effort setting is cheapest?
In Artificial Analysis’s evaluation, low effort averaged $0.055 per task, compared with $0.107 for non-reasoning and $0.40 at maximum effort. The results are averages from that benchmark and may not predict costs for an individual workload.
Are Terra’s speed and safety claims independently verified?
The source material attributes the claim that Terra is more than three times faster than Sol and Astra, and its cybersecurity classification, to OpenAI. It says Artificial Analysis had not published speed measurements when its launch-day evaluation was checked; independent verification of those particular claims was not provided.
Source: ThorstenMeyerAI.com
