🔍 Read the full analysis: Does GPT-6.1 Terra Offer The Right Balance Of Cost And Capability? 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 Oct. 26, 2026, pricing API use at $1 per million input tokens and $5 per million output tokens. Artificial Analysis measured its task cost at $0.40 on max effort and found low and medium settings on its cost-efficiency frontier; Terra still scored below several higher-capability models, and the source material does not provide the complete six-setting results.
OpenAI released GPT-6.1 Terra on Oct. 26, positioning it as a lower-cost model for workplace and computer-based tasks, with API prices of $1 per million input tokens and $5 per million output tokens. Artificial Analysis reported that Terra’s max-effort setting averaged $0.40 per task and scored 48.1 on its Intelligence Index, while its lower and medium effort settings offered a stronger cost-to-score balance on that benchmark.
OpenAI lists Terra in ChatGPT Work, Codex and its API under the model identifier gpt-6.1-terra. The stated API rates are half the listed price of its predecessor, GPT-5.6 Terra, at $2 per million input tokens and $12 per million output tokens. Cached input is priced at $0.05 per million tokens. OpenAI says eligible Plus, Pro, Enterprise and Edu users can use Terra in ChatGPT Work and Codex, and that the same allowance can cover substantially more tasks than with Sol or Astra.
Artificial Analysis’ launch-day evaluation puts Terra at 48.1 on its Intelligence Index at max effort, at an average cost of $0.40 per task across the index’s evaluations. In the same comparison, GPT-6.1 Sol scored 51.8 at $0.72 per task, Claude Opus 5.5 scored 57.6 at $5.98, and GPT-6 Luna scored 38.1 at $0.07. These figures measure different models’ results and average evaluation-task costs; they do not establish what any particular customer will pay for a real-world job.
OpenAI’s launch-page figures also report gains on selected tests. At medium effort on Agents’ Last Exam, Terra scored 57.6%, compared with 42.6% for GPT-5.6 Terra, and OpenAI listed costs of $0.29 and $0.84 per task, respectively. On Terminal-Bench 4.0 at max effort, OpenAI reported Terra solved 50.3% of tasks, against 21.5% for its predecessor. These are company-reported benchmark results, distinct from Artificial Analysis’ independent index.
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 a Finished Task
The release highlights why token prices alone can mislead. A model with cheaper tokens may produce more output or use more reasoning, while a pricier model may complete a job with fewer tokens. Artificial Analysis found that Luna, listed at $0.10 per million input and output tokens, averaged $0.07 per evaluation task at max effort; Terra’s max setting averaged $0.40. However, the source says Terra at low effort averaged $0.055 per task—below Luna’s reported average—illustrating how task-level comparisons can differ from headline token rates.
For organizations processing invoices, handling routine computer work or operating coding agents, per-task cost can help with budgeting and model routing. A low-cost setting may be suitable for repetitive work, with a more capable model reserved for harder tasks. The trade-off is that lower benchmark scores may matter when a task requires complex reasoning, accuracy or a reliable answer on the first attempt. Cost and capability need to be judged against the specific workload, not a single overall ranking.
Terra is not the highest-scoring model in the figures supplied. Artificial Analysis placed Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon above it on the Intelligence Index. It also reported Xiaomi’s open-weight MiMo-V2.6-Pro at 46.3 for $0.13 per task. That comparison makes Terra’s appeal a price-and-performance position, not a claim that it leads the market.
Products Worth Considering

LG UltraGear 32GX870A-B 32-inch 4K UHD OLED Dual-Mode Gaming Monitor

Samsung 27-Inch Odyssey OLED G6 QHD Gaming Monitor (G61SH)

LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1, Black
As an affiliate, we earn on qualifying purchases.
Where Terra Fits in OpenAI’s Lineup
OpenAI describes Terra as a fast, efficient model for communication and computer-based work, citing invoice reading and entry into accounting software as an example. The company places it between Luna, its smaller model, and Sol and Astra. OpenAI says Terra completes tasks more than three times faster than Sol and Astra, but Artificial Analysis had not published speed measurements for Terra when the source article checked on Oct. 26. The speed statement is therefore an OpenAI claim, not an independently confirmed result in the cited material.
Artificial Analysis says its Intelligence Index combines ten evaluations, including agentic office work, terminal tasks, science and hallucination testing, and records associated costs including cached input and reasoning tokens. Its figures are an average across those evaluation tasks, not a universal price quote. The source also reports that Terra at max effort was six index points above GPT-5.6 Terra for less than a third of its predecessor’s average task cost, according to Artificial Analysis.
The available source describes six effort settings, from non-reasoning to max, and says cost and score rise across the curve. It reports that max costs 7.4 times as much as low for 12 additional index points, and that low and medium lie on Artificial Analysis’ cost-efficiency frontier. The supplied material gives only part of the setting-by-setting table, so the full figures cannot be independently restated here.
Limits of the Launch-Day Evidence
The benchmark results do not establish how Terra will perform across every customer’s software, prompts or workload. Artificial Analysis’ figures are averages over its own evaluation set, while OpenAI’s computer-use, coding and factual-error results come from company-reported tests. The provided material does not include the full six-setting table, so exact scores and costs for every effort level cannot be checked from it.
Other evidence is also incomplete. Artificial Analysis had not published speed measurements at the time of the report, leaving OpenAI’s claim that Terra is more than three times faster than Sol and Astra without independent confirmation in the cited material. OpenAI also said Terra meets the “Critical” threshold in cybersecurity under its Preparedness Framework; that classification is the company’s own assessment. The materials do not specify how actual usage costs will vary by task, nor do they provide independent confirmation of the claimed allowance gains.
Workload Tests Will Settle the Trade-Off
Terra is available now through the listed ChatGPT Work, Codex and API channels, so the next practical test is how its cost and reliability compare on real work. Customers evaluating it can measure cost per successfully completed task, error rates, speed and the amount of human correction required, rather than relying on token prices or a single benchmark score.
Further independent speed measurements and broader evaluations across Terra’s effort settings would help clarify where it fits against Luna, Sol, Astra and competing models. For now, the evidence supports a narrower conclusion: Terra offers lower listed token rates than its predecessor and has competitive task-cost results at some effort levels, but buyers must weigh those savings against the capability required for each job.
Key Questions
How much does GPT-6.1 Terra cost?
OpenAI lists API prices of $1 per million input tokens and $5 per million output tokens. Cached input is listed at $0.05 per million tokens. Actual cost depends on usage and task.
Is Terra the highest-scoring model in the comparison?
No. Artificial Analysis gave max-effort Terra an Intelligence Index score of 48.1. The supplied comparison lists higher scores for Claude Opus 5.5, Kimi K3.1 Pro, Gemini 4 Argon, GPT-6 Astra and GPT-6.1 Sol.
Why compare cost per task instead of token price?
Token rates do not show how many tokens a model uses to finish a job. Task-level costs can reflect input, output and reasoning usage, though Artificial Analysis’ averages may not match a particular customer’s workload.
Which Terra effort setting should users choose?
The source reports that low and medium settings sit on Artificial Analysis’ cost-efficiency frontier, but it does not provide the complete setting-by-setting results. The appropriate setting depends on the task’s difficulty and the acceptable balance between cost and performance.
Source: ThorstenMeyerAI.com
