GPT-6.1 Terra Could Be OpenAI’s Everyday Value Option
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: GPT-6.1 Terra Could Be OpenAI’s Everyday Value Option on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI released GPT-6.1 Terra on October 26, 2026, positioning it as a lower-priced model for routine work across ChatGPT Work, Codex and its API. Artificial Analysis found that Terra’s low and medium reasoning settings offer a strong cost-to-score balance on its tests, but higher effort costs more, and leading models score better.

OpenAI released GPT-6.1 Terra on October 26, 2026, adding a lower-priced model for routine work to ChatGPT Work, Codex and its API. Its listed API rates are $1 per million input tokens and $5 per million output tokens, half the prices cited for its predecessor, GPT-5.6 Terra; independent testing suggests Terra’s value depends on how much reasoning a task needs.

OpenAI describes Terra as a “fast, efficient model” for communication and computer-based work, including handling invoices and entering information into accounting software. The model is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and through the API as gpt-6.1-terra. OpenAI says a given allowance can cover substantially more tasks with Terra than with Sol or Astra. Its launch materials also list cached input at $0.05 per million tokens.

Artificial Analysis, a benchmarking firm, published an independent evaluation on launch day. It gave Terra a score of 48.1 on its Intelligence Index at maximum effort, compared with 51.8 for GPT-6.1 Sol and 57.6 for Claude Opus 5.5. Its reported average cost per task was $0.40 for Terra at maximum effort, versus $0.72 for Sol and $5.98 for Opus 5.5. Those figures reflect the benchmark’s evaluation tasks, not a guaranteed cost for every customer workload.

The evaluator’s six-setting results show why token prices alone can mislead. Terra’s low-effort setting scored 36.2 at an average cost of $0.055 per task; the non-reasoning setting cost $0.107 and scored 33.5. At maximum effort, the score rose to 48.1, while average task cost reached $0.40. Artificial Analysis places low and medium effort on its cost-efficiency frontier; the higher settings do not offer the same score-for-cost position in that comparison.

At a glance
announcementWhen: Announced October 26, 2026
The developmentOpenAI released GPT-6.1 Terra, a model priced below GPT-6.1 Sol, alongside launch-day benchmark and cost-per-task comparisons.
Thorsten Meyer AI
Artificial Analysis · 26 Oct 2026

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.

$0.055per task at low effort, the cheapest Terra setting
7.4×the cost of max over low, for 11.9 more index points
2.1×the cost of max over xhigh, for 2.7 more points
2×non-reasoning costs about twice as much as low, and scores lower

Score · Intelligence Index

independent higher is better

  • Non-reasoning33.5
  • Lowfrontier36.2
  • Mediumfrontier41.3
  • High43.9
  • Xhigh45.4
  • Max48.1

Cost per finished task

note the last step shorter is cheaper

  • Non-reasoning4,737 output tokens$0.107
  • Low2,681 output tokens$0.055
  • Medium5,835 output tokens$0.095
  • High9,027 output tokens$0.134
  • Xhigh14,551 output tokens$0.194
  • Max35,096 output tokens$0.403

All six settings

EffortIndexCost / taskOutput tokens / taskHallucination rate
Non-reasoningIndex 33.5Cost $0.107Tokens 4,737Hallucination 72%
LowIndex 36.2Cost $0.055Tokens 2,681Hallucination 46%
MediumIndex 41.3Cost $0.095Tokens 5,835Hallucination 44%
HighIndex 43.9Cost $0.134Tokens 9,027Hallucination 43%
XhighIndex 45.4Cost $0.194Tokens 14,551Hallucination 42%
MaxIndex 48.1Cost $0.403Tokens 35,096Hallucination 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 Completing Each Task

Terra’s release reflects a practical choice for organisations that run many routine AI tasks: a model’s price per token does not by itself show how much it costs to finish a job. A cheaper model may generate more tokens, take more steps or need retries. Artificial Analysis’s task-cost measure accounts for spending across its benchmark, making it a more useful comparison for that test set than the headline token rate alone.

For businesses handling repetitive work, the reported low- and medium-effort results could make Terra worth evaluating for high-volume workflows. But the numbers do not establish savings for a particular company. Actual cost will depend on task complexity, usage patterns, model settings and whether the output meets the required standard without extra review. OpenAI says Terra is faster than Sol and Astra, but Artificial Analysis had not published speed measurements for Terra at the time of its launch-day comparison.

The trade-off is capability. Terra’s maximum-effort score trails several models in Artificial Analysis’s table, including Opus 5.5, Kimi K3.1 Pro and Gemini 4 Argon. The benchmark therefore supports a narrower interpretation: Terra may offer a useful balance on some workloads, rather than matching the strongest models across the board.

Products Worth Considering

Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, White Smoke

Bose QuietComfort Headphones (2nd Gen), Wireless Noise-Cancelling Over-Ear Headphones, White Smoke

Bose QuietComfort Wireless Noise-Cancelling Headphones, Moonlight Grey

Bose QuietComfort Wireless Noise-Cancelling Headphones, Moonlight Grey

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptics – Graphite

Logitech MX Master 4 Ergonomic Wireless Mouse with Haptics – Graphite

As an affiliate, we earn on qualifying purchases.

Where Terra Fits in OpenAI’s Lineup

OpenAI positions Terra between its smaller GPT-6 Luna and its larger Sol and Astra models. The source article describes Terra’s API rates as half those of GPT-5.6 Terra, while the listed rates for GPT-6.1 Sol are $2 per million input tokens and $10 per million output tokens. OpenAI’s stated use case is everyday communication and computer-based work; it says Sol and Astra remain better choices for the most complex coding tasks.

On Artificial Analysis’s index, Luna scored 38.1 at maximum effort and averaged $0.07 per task, while Terra at low effort scored 36.2 and averaged $0.055. That comparison illustrates why a model with a higher token price can still cost less for a completed benchmark task. The figures should be read as results from a defined test suite, not universal rankings or predictions for every user.

Other launch-day figures come from OpenAI’s own tests and should be distinguished from the independent evaluation. OpenAI reported that Terra at medium effort scored 57.6% on Agents’ Last Exam for $0.29 per task, compared with 42.6% and $0.84 for GPT-5.6 Terra. It also reported a 50.3% score on Terminal-Bench 4.0 at maximum effort, compared with a previously published 21.5% for GPT-5.6 Terra. These results are vendor-reported and may use different methods from third-party benchmarks.

Limits of the Launch-Day Evidence

It is not yet clear how Terra will perform or price out across real customer workloads. Artificial Analysis reports averages across its own tasks, and OpenAI’s separate benchmark results are company-reported; neither set of figures establishes typical savings for all users. The source material also does not provide a full breakdown of Terra’s measured results at every effort setting, so the available figures do not support reconstructing the entire cost-and-score curve.

Some comparisons involve different evaluations and should not be treated as direct head-to-head results. The source gives OpenAI’s speed claim but says independent speed measurements were not available when Artificial Analysis published its evaluation. OpenAI’s statement that Terra meets the “Critical” threshold in cybersecurity under its Preparedness Framework is also the company’s own classification; the cited material does not include an outside assessment.

Benchmark scores cannot establish whether a model is suitable for a specific business process, how often human checking will be needed, or what the total cost of deployment will be. Those questions require testing with the organisation’s own data and workflow.

Testing Terra on Real Workloads

The immediate next step for prospective users is to compare Terra with their current model on representative tasks, tracking cost per accepted result, error rates and review time—not token price alone. Organisations can start with low or medium effort for routine work, then test whether the added expense of higher settings improves results enough to justify it. For difficult coding or other demanding tasks, OpenAI itself points users toward Sol or Astra.

Further independent evaluations could clarify Terra’s speed, performance across effort settings and reliability outside the launch-day benchmark suite. OpenAI’s pricing and availability are confirmed in the supplied launch information; broader adoption, customer savings and longer-term performance remain to be established.

Key Questions

What is GPT-6.1 Terra?

GPT-6.1 Terra is an OpenAI model released on October 26, 2026, for communication and computer-based tasks. It is available through ChatGPT Work, Codex and the API.

How much does GPT-6.1 Terra cost?

OpenAI’s listed API prices are $1 per million input tokens and $5 per million output tokens, with cached input listed at $0.05 per million tokens. A benchmark’s average cost per task is a separate measure and can vary with the task and reasoning setting.

Is Terra cheaper than GPT-6 Luna?

Terra’s listed price per token is higher than Luna’s, but Artificial Analysis reported that Terra at low effort cost less per task on average than Luna at maximum effort in its evaluation. The comparison is specific to that benchmark; it does not establish which model will be cheaper for every workload.

Does Terra match OpenAI’s larger models?

Not across the cited results. Artificial Analysis scored Terra below GPT-6.1 Sol and several other models on its Intelligence Index. OpenAI says Sol and Astra remain better choices for its most complex coding tasks.

Which Terra reasoning setting should users choose?

The source’s benchmark findings favor low or medium effort when balancing score and task cost, but the right setting depends on the work. Users should test whether higher effort improves accuracy enough to warrant its higher measured cost.

Source: ThorstenMeyerAI.com

You May Also Like

Should You Set The Shared AI Training Opt-Out? What It Covers

OpenAI, Google, Microsoft and Anthropic say they will honor a shared robots.txt training preference, but rollout dates and coverage differ.

Master Relationship Building With Pre-Call Memory Cards In CRM Systems

Testing of pre-call memory cards for relationship-driven professionals aims to improve client engagement by capturing human context in CRMs, using AI summarization.

AI For Science Expands To Foundations And Research Institutions – Anthropic

Anthropic says its AI for Science initiative will expand to foundations and research institutions, but has not named partners or outlined the program.

AI SRE: What Is It, How Could It Change On-call – And Does It Work?

AI SRE tools are gaining traction as incident pages rise, but current systems focus on investigation and triage, not unsupervised production fixes.