🔍 Read the full analysis: GPT-6.1 Terra: A Possible Shortcut To Lower AI Spending 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 Oct. 26, 2026, positioning it as a lower-cost model for routine work. Artificial Analysis measured a $0.40 average cost per evaluation task at maximum effort and found low and medium settings on its cost-efficiency frontier; the model does not lead the benchmark, and some figures come from OpenAI’s own tests.
OpenAI lists API pricing at $1 per million input tokens and $5 per million output tokens, compared with $2 and $12 for GPT-5.6 Terra. Cached input is priced at $0.05 per million tokens. The model is available through the API as gpt-6.1-terra and to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. OpenAI says the same allowance can cover “substantially more” tasks than with Sol or Astra.
Artificial Analysis gave Terra a 48.1 Intelligence Index score at maximum effort, with an average cost of $0.40 per task across its 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. The index combines ten evaluations, including office work, terminal tasks, science and hallucination. These are benchmark results, not guarantees of what a particular user will pay or achieve.
In Artificial Analysis’s six-setting results, Terra’s score ranged from 33.5 at non-reasoning to 48.1 at maximum effort, while average task cost ranged from $0.055 at low effort to $0.40 at maximum. Low and medium were on the firm’s cost-efficiency frontier; settings above those levels were not. The supplied launch-day material does not give the full figures for every setting, so the relationship should not be read as a universal cost estimate for every workload.
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 Per-Task Costs Matter
The launch highlights a practical distinction for organizations buying AI: token prices do not determine the full cost of completing work. A model with a cheaper token rate can use more tokens or take a longer route to finish a task. Artificial Analysis reported that GPT-6 Luna is ten times cheaper than Terra per token, yet Terra at low effort costs less per task in its measurements because it produces fewer tokens.
That makes Terra a possible fit for high-volume, repeatable work, such as the invoice processing example OpenAI cited. But the benchmark does not establish that Terra will be cheaper for every business process. Teams would need to test their own tasks, including error correction and human review, and compare total spending per successfully completed task. For difficult coding or reasoning work, a higher-capability model may still be needed.
The performance trade-off is visible in the scores. At maximum effort, Terra’s Intelligence Index result is below Sol and several other models in the table. Artificial Analysis also listed open-weight MiMo-V2.6-Pro at a score of 46.3 and $0.13 per task, below Terra’s maximum-effort cost and score. The figures offer a benchmark comparison, not a direct measure of service quality across all uses.
Products Worth Considering
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Terra’s Place in OpenAI’s Lineup
OpenAI describes Terra as a fast, efficient model for communication and computer-based work. Its launch example is reading incoming invoices and entering information into accounting software. The model sits between the smaller Luna and larger Sol and Astra models in the lineup, according to the source material. OpenAI says Terra completes tasks more than three times faster than Astra and Sol, but Artificial Analysis had not published speed measurements when the launch-day report was prepared.
OpenAI also published its own task-specific results, which should be distinguished from the independent index. On Agents’ Last Exam, OpenAI reported Terra at medium effort scoring 57.6%, compared with 42.6% for GPT-5.6 Terra, at reported costs of $0.29 and $0.84 per task respectively. At maximum effort, OpenAI reported a $1.01 cost for Terra against $8.82 for Claude Opus 5.5 on that test, while Opus scored 63.2% and Terra 60.0%.
Other company-reported results include a 50.3% score on Terminal-Bench 4.0 at maximum effort, compared with 21.5% for GPT-5.6 Terra. OpenAI said its internal error-inducing prompt set showed error rates falling from 53.6% to 16.7% at low effort and from 22.3% to 4.3% at maximum effort. These are OpenAI’s measurements and are not the same as independent benchmark findings.
Limits of the Cost Comparisons
The $0.40 per-task figure is an average across Artificial Analysis’s evaluation tasks, not a quoted price for an individual user’s workflow. Actual costs can vary with task length, model setting, retries, cached input and the amount of human review required. The source material does not provide full results for every effort setting or enough detail to calculate costs for specific business deployments.
Several comparisons are based on different tests: Artificial Analysis’s index, OpenAI’s own benchmarks and OpenAI’s internal error prompts. Their scores and costs should not be treated as directly interchangeable. Artificial Analysis had not published speed measurements for Terra at the time of the report, and OpenAI’s speed claim was not independently verified there. The source also says Terra guessed on 38% of questions it could not answer in the firm’s hallucination test, down from 88% for GPT-5.6 Terra, but does not supply the test’s full methodological detail here.
OpenAI classifies Terra as meeting the “Critical” threshold in cybersecurity under its Preparedness Framework and says it retains safeguards developed for GPT-6 Astra. That classification is the company’s own. The launch material does not establish how Terra will perform across real-world deployments or whether its benchmark advantages will persist on different workloads.
Testing Terra on Real Workloads
The immediate next step for prospective users is to compare Terra with alternatives on their own tasks, recording cost per successfully completed task alongside accuracy, speed and any correction work. For routine processes, the benchmark findings point to low or medium effort as sensible starting settings, rather than assuming maximum effort is automatically worth its added cost.
Companies should also test when to route work to Luna, Terra, Sol or Astra. OpenAI itself says Sol and Astra remain better choices for its most complex coding tasks. Further independent speed measurements, broader deployment data and task-specific evaluations would help clarify whether Terra’s lower listed prices translate into lower total operating costs. No such results were supplied in the launch-day material.
Key Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is a model OpenAI released on Oct. 26, 2026, aimed at communication and computer-based tasks. It is available through the API, ChatGPT Work and Codex for listed paid plans.
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 $0.40 per evaluation task at maximum effort; that is a benchmark average, not a fixed price for every task.
Is Terra cheaper than GPT-6 Luna?
Terra is more expensive per token than Luna in the supplied comparison, but Artificial Analysis found Terra at low effort cost less per task in its evaluation because it used fewer tokens. Results may differ on other workloads.
Does Terra outperform OpenAI’s larger models?
Not across the cited independent index. At maximum effort, Terra scored 48.1, below GPT-6.1 Sol at 51.8. OpenAI also says Sol and Astra remain better choices for its most complex coding tasks.
Which effort setting should users choose?
Artificial Analysis placed low and medium effort on its cost-efficiency frontier. Users should test those settings against their own accuracy and cost requirements; the benchmark does not establish one best setting for every task.
Source: ThorstenMeyerAI.com
