🔍 Read the full analysis: GPT-6.1 Terra May Offer A Cheaper Route To Everyday AI Work on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
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 measured its cost per task at $0.055 for low effort and $0.40 at maximum effort; the cheaper settings can be more cost-efficient, but Terra does not match higher-scoring models on the reported index.
OpenAI released GPT-6.1 Terra on October 26, adding a lower-priced model for everyday work in ChatGPT Work, Codex and its API. The model costs $1 per million input tokens and $5 per million output tokens; independent testing by Artificial Analysis suggests its low- and medium-effort settings may deliver a comparatively inexpensive way to complete benchmark tasks, though higher-scoring models remain available.
OpenAI describes Terra as a fast, efficient model for communication and computer-based work, such as reading invoices and entering information into accounting software. Its listed API rates are half the rates given for GPT-5.6 Terra: $1/$5 per million input/output tokens, compared with $2/$12. Cached input is priced at $0.05 per million tokens. OpenAI says eligible Plus, Pro, Enterprise and Edu users can access Terra in ChatGPT Work and Codex, where the same allowance can cover more tasks than with Sol or Astra.
Artificial Analysis published launch-day measurements across six effort settings. In its Intelligence Index, Terra scored 48.1 at maximum effort and cost an average of $0.40 per task across the evaluated tasks. GPT-6.1 Sol scored 51.8 at $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. These are benchmark results, not a guarantee of what a particular user’s task will cost or how well the model will perform.
The reported effort curve shows a trade-off between score and spending: Terra ranged from 33.5 points and $0.107 per task in non-reasoning mode to 48.1 points and $0.40 at maximum effort. Low effort scored 36.2 for $0.055 per task. Artificial Analysis describes low and medium settings as being on its cost-efficiency frontier. Its supplied figures also show that non-reasoning mode cost nearly twice as much per task as low effort while scoring lower, underscoring that token rates alone do not determine task cost.
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 Calculation
Terra’s launch puts attention on cost per completed task, rather than price per token alone. A model with a low token rate can still spend more if it uses many tokens to complete a task; conversely, a model that produces fewer tokens may cost less overall. Artificial Analysis’s measurements indicate that Terra at low effort can undercut larger models on its benchmark’s average task cost, while also scoring below them. For businesses processing high volumes of routine work, that trade-off could affect which model they assign to each job.
The practical implication is not that Terra replaces a higher-capability system across the board. The available results point instead to a possible tiered model strategy: use a less expensive setting for routine requests and reserve higher-effort or larger models for work where performance matters more than cost. OpenAI itself says Sol and Astra remain better choices for its most complex coding tasks. Actual savings will depend on the task mix, token use, model settings and the price terms customers receive.
Popular right now
Philips 4400 Series Fully Automatic Espresso Machine with LatteGo Milk System
eufy E25 Robot Vacuum and Mop Combo with HydroJet Roller Mop
Ninja Foodi 10-Qt DualZone Smart XL Air Fryer, 6-in-1, DZ550
As an affiliate, we earn on qualifying purchases.
Terra Between Luna and Sol
OpenAI positions Terra between its smaller Luna model and larger Sol and Astra models. The launch follows GPT-5.6 Terra, and Artificial Analysis says the new version gains six points on its Intelligence Index at maximum effort while costing less than a third as much per average benchmark task: $0.40 versus $1.40. That comparison comes from the benchmark firm’s reported measurements, not a universal price for every use.
The same independent index places Terra below several higher-scoring models, including Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon. It also lists Xiaomi’s open-weight MiMo-V2.6-Pro at 46.3 points and $0.13 per task, a lower cost and a higher score than Terra’s high and extra-high settings, according to the supplied comparison. OpenAI separately reported results for computer use and coding, including a 50.3% score for Terra at maximum effort on Terminal-Bench 4.0, against 21.5% for GPT-5.6 Terra. Those figures are OpenAI’s own benchmark claims and should not be treated as independent measurements.
Limits of the Launch-Day Evidence
Benchmark averages do not settle real-world performance. Artificial Analysis’s task-cost figure averages spending across its own evaluation tasks, and costs for a specific workflow may differ with prompt length, output length, cached input and reasoning setting. The provided source does not include the complete figures for every point on Terra’s six-setting curve, so comparisons beyond the listed settings cannot be fully checked here.
OpenAI’s claims about speed, computer use, coding, factual errors and cybersecurity are company-reported results or classifications. The supplied material says Artificial Analysis had not published speed measurements for Terra at the time of its review. It also does not establish how Terra will perform across independent production deployments, how availability may vary by plan or region, or whether prices will change. The launch-day data therefore supports a measured comparison, not a settled judgment about the best model for every job.
Watch Real-World Costs and Access
Terra is available through the named ChatGPT plans, Codex and the API, according to the launch information. The next useful evidence will be independent testing across more tasks, including speed measurements and performance in routine business workflows, alongside clearer comparisons of the effort settings. Customers considering a switch can compare actual spending and output quality on their own representative tasks rather than relying only on token prices or benchmark averages.
OpenAI has not provided a further launch milestone in the source material. For now, the main open question is whether Terra’s lower listed rates and benchmark task costs translate into reliable savings for customers without requiring more review or correction. That will depend on how well the model handles each organization’s work and how often users need to escalate difficult jobs to Sol, Astra or another higher-scoring model.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model aimed at communication and computer-based work. OpenAI says it is available in ChatGPT Work, Codex and the API.
How much does Terra cost?
OpenAI lists API pricing at $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. These token rates are separate from the benchmark’s average cost-per-task estimates.
Is Terra cheaper than GPT-6.1 Sol?
Artificial Analysis measured Terra at $0.40 per average task at maximum effort, compared with $0.72 for Sol. Its figures depend on the benchmark tasks and do not establish what every real-world request will cost.
Does Terra outperform larger models?
Not on the reported Artificial Analysis Intelligence Index: Terra scored 48.1 at maximum effort, below Sol at 51.8 and Opus 5.5 at 57.6. OpenAI says Sol and Astra remain better choices for its most complex coding tasks.
Which Terra setting is the cheapest?
Among the supplied figures, low effort cost $0.055 per average task, less than non-reasoning mode at $0.107. It also scored higher on the index: 36.2 compared with 33.5.
Source: ThorstenMeyerAI.com
