🔍 Read the full analysis: Is GPT-6.1 Terra The Cheapest Way To Get The Job Done? on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI released GPT-6.1 Terra on October 26, 2026, positioning it as a lower-cost model for routine workplace and computer-based tasks. Artificial Analysis reported a $0.40 average task cost and an Intelligence Index score of 48.1 at maximum effort; its figures also show that low and medium settings can offer better score-per-dollar trade-offs, while other models may cost less or score higher.
OpenAI released GPT-6.1 Terra on October 26, describing it as a fast, efficient model for workplace communication and computer-based tasks. Its API price is $1 per million input tokens and $5 per million output tokens, half the listed per-token price of its predecessor, GPT-5.6 Terra; independent testing puts Terra’s average cost at $0.40 per task at maximum effort, but does not make it the top-scoring or cheapest model overall.
Terra is positioned between OpenAI’s smaller GPT-6 Luna and 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 name gpt-6.1-terra. The company says the same ChatGPT allowance can cover substantially more tasks with Terra than with Sol or Astra. That is OpenAI’s description, not an independent measure of how far an individual user’s allowance will go.
Artificial Analysis, an independent benchmarking firm, scored Terra at 48.1 on its Intelligence Index at maximum effort, with an average evaluation-task cost of $0.40. In the same comparison, GPT-6.1 Sol scored 51.8 at $0.72 per task, while Claude Opus 5.5 scored 57.6 at $5.98. Those figures describe average cost across the benchmark’s evaluation tasks, not a guaranteed charge for every real-world job.
The comparison also changes with Terra’s reasoning effort. Artificial Analysis reported six settings ranging from a non-reasoning score of 33.5 at $0.107 per task to a maximum-effort score of 48.1 at $0.40. Low effort scored 36.2 at an average $0.055 per task. The supplied benchmark summary says low and medium settings sit on the cost-efficiency frontier; higher settings cost more for their incremental scores. It also reports that non-reasoning mode costs nearly twice as much per task as low effort, despite 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
Why Task Cost Changes the Choice
Terra’s launch highlights a practical distinction for businesses buying AI services: token price is not the same as task cost. A model can charge less for each token but use more tokens to complete a job. Artificial Analysis reports that Luna’s listed token rates are lower, yet Terra at low effort costs less per evaluated task because it produces fewer tokens in those tests. That result is specific to the benchmark and should not be treated as a universal outcome for every prompt or workflow.
For organizations processing large volumes of routine work, average cost per completed task can be more useful for budgeting than a rate card alone. The source gives invoice reading and entering data into accounting software as an example of the kind of high-volume work OpenAI has in mind. At the same time, benchmark scores and costs do not by themselves establish reliability, integration costs, or the amount of human review a production workflow needs. A cheaper model is only a better fit if its output meets the task’s requirements.
Terra’s results also show the trade-off between expenditure and capability. At maximum effort, it costs less per benchmark task than Sol or Opus 5.5, but scores below both. OpenAI says Sol and Astra remain better choices for the most complex coding work. For users and companies, the choice is not simply “cheapest model”: it is which model and effort level can complete a particular job to an acceptable standard at an acceptable total cost.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Terra’s Place in OpenAI’s Lineup
OpenAI describes Terra as a model for the communication and computer-based work that fills a typical workday. It lists a workflow involving reading incoming invoices and entering them into accounting software as an example. The company says Terra’s API prices are 50% lower than GPT-5.6 Terra’s, at $1/$5 per million input/output tokens, compared with $2/$12 for the earlier model. Cached input is listed at $0.05 per million tokens.
OpenAI’s launch materials include its own results alongside the independent evaluation. On Agents’ Last Exam, OpenAI says Terra at medium effort scored 57.6% and cost $0.29 per task, compared with $0.84 for GPT-5.6 Terra at medium effort. On Terminal-Bench 4.0, the company reports a 50.3% solve rate at maximum effort for Terra, versus 21.5% for GPT-5.6 Terra’s published result. Those are company-reported benchmark figures; the tasks and conditions differ by evaluation, so they should not be combined into a single general performance claim.
Artificial Analysis’ Intelligence Index combines ten evaluations covering areas including agentic office work, terminal tasks, science and hallucination. Its figures, retrieved October 26, 2026, place Terra below several higher-scoring models and above others. Xiaomi’s open-weight MiMo-V2.6-Pro, for example, scored 46.3 at an average $0.13 per task in the supplied comparison, below Terra’s score but at a lower cost. The table is a dated snapshot, and its results depend on the selected tests and model settings.
Limits of the Cost Comparison
The benchmark’s average cost per task is not a quote for a specific customer workflow. Actual charges can vary with prompt length, output length, cached input, reasoning effort, tool use and the number of attempts needed. The supplied source does not provide a real-world cost study showing how Terra performs across customers’ production tasks.
Some measures cited at launch come from OpenAI, while others come from Artificial Analysis, and they are not interchangeable. OpenAI’s reported speed advantage—more than three times faster than Astra and Sol—had no corresponding Terra speed measurements from Artificial Analysis at the time reported. OpenAI also classifies Terra as meeting the “Critical” threshold in cybersecurity under its Preparedness Framework; that is the company’s classification, not an independent finding presented here.
The supplied material lists Terra’s hallucination test result as a reduction in guessing on questions it could not answer, from 88% for GPT-5.6 Terra to 38% for GPT-6.1 Terra. It does not provide enough detail here to establish how that test maps to error rates in ordinary use. Results can also change with the effort setting: the source reports a 72% hallucination rate for non-reasoning mode and 46% at low effort. The figures should be read with their test conditions, which are not fully described in the material provided.
Testing Terra in Real Workflows
Terra is available through the API, ChatGPT Work and Codex, according to OpenAI, so the next evidence will come from users applying it to actual workflows and from further independent evaluations. Companies considering adoption can compare settings on representative tasks, measuring cost per accepted result as well as speed, error rates and the need for human checking. The launch-day benchmark suggests starting with low or medium effort for cost-sensitive tasks, then testing whether more effort produces enough additional quality to justify its added cost.
Future comparisons may also clarify how Terra’s speed, coding performance and error rates hold up under consistent test conditions across competing models. Until those results are available, the published figures support a narrower conclusion: Terra offers a lower per-token price than GPT-5.6 Terra and a relatively low measured task cost, but whether it is the cheapest suitable option depends on the job, the effort setting and the standard of accuracy required.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a new OpenAI model positioned between Luna and the larger Sol and Astra models. OpenAI says it is designed for communication and computer-based workplace tasks and lists it in ChatGPT Work, Codex and the API.
What does GPT-6.1 Terra cost?
OpenAI lists API rates of $1 per million input tokens and $5 per million output tokens, with cached input at $0.05 per million tokens. Those are token prices; a real task’s cost depends on how many tokens and other resources it uses.
Is Terra the cheapest model for completing tasks?
Not across the full comparison supplied. Artificial Analysis measured Terra at $0.40 per task at maximum effort, while MiMo-V2.6-Pro averaged $0.13 and Luna $0.07. Terra’s low-effort result was $0.055 per task, but comparisons depend on model settings and benchmark conditions.
Which Terra effort setting offers the best value?
Artificial Analysis’ supplied summary places low and medium effort on the cost-efficiency frontier. Low effort scored 36.2 at an average $0.055 per task. The best setting for a particular use depends on whether its output quality is adequate for that job.
Does Terra outperform OpenAI’s larger models?
Not on the cited maximum-effort Intelligence Index results. Terra scored 48.1, compared with 51.8 for GPT-6.1 Sol and 52.7 for GPT-6 Astra. OpenAI also says Sol and Astra remain better choices for the most complex coding tasks.
Source: ThorstenMeyerAI.com
