GPT-6.1 Terra: OpenAI's Mid-Tier Model Pricing, Benchmarks, and HN Debate

GPT-6.1 Terra is OpenAI's new mid-tier model, positioned as a cheaper and faster alternative to GPT-6.1 Sol and GPT-6 Astra while improving on GPT-5.6 Terra in visual reasoning, computer use, factuality, and coding. The release matters because OpenAI is selling it primarily on cost per task: Terra is priced at $1 per million input tokens and $5 per million output tokens, with OpenAI claiming large efficiency gains on professional computer-use and coding benchmarks.

What GPT-6.1 Terra is and what it costs

GPT-6.1 Terra is available to all Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex, and developers can access it through the OpenAI API as gpt-6.1-terra. Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens.

OpenAI says Terra's standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol. A Hacker News commenter summarized the resulting API price ladder as GPT-6 Astra at $10 input / $50 output per million tokens, GPT-6.1 Sol at $2 / $10, GPT-6.1 Terra at $1 / $5, and GPT-6 Luna at $0.10 / $0.50.

OpenAI describes Terra as continuing its cost-intelligence frontier across model families. The company says the lower price makes Terra practical for high-volume automation such as reading incoming invoices and entering their details into accounting software. In ChatGPT Work and Codex, OpenAI says the same usage allowance covers more tasks with GPT-6.1 Terra than with GPT-5.6 Terra and substantially more than with GPT-6.1 Sol or GPT-6 Astra.

Benchmarks: computer use, visual reasoning, factuality, and coding

OpenAI's benchmark claims center on cost per task rather than top-end capability. On Agents' Last Exam, a benchmark of professional tasks that includes turning architectural drawings into a 3D model and reviewing an animated film shot by shot, GPT-6.1 Terra at medium reasoning effort scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost. OpenAI says Terra performs on par with GPT-6 Astra at only one-tenth its price per token. At max effort, OpenAI says Terra is over three times as token efficient as Claude Opus 5.5 and one-fourth its price per token, resulting in roughly 9x lower cost per task. OpenAI also says Terra completes tasks over three times faster than both GPT-6 Astra and GPT-6.1 Sol.

On MMMU-Pro, a benchmark of visual reasoning, GPT-6.1 Terra outperforms GPT-5.6 Terra at every reasoning setting. OpenAI says Terra nearly matches the scores of GPT-6 Astra and GPT-6.1 Sol at much lower cost. On OSWorld 2.1, which tests how well AI agents use desktop apps to complete multistep tasks, GPT-6.1 Terra at medium reasoning effort surpasses GPT-5.6 Terra's best score at a fraction of the cost per task.

On factuality, OpenAI says its internal benchmark of error-inducing prompts showed the error rate at low reasoning effort falling from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra, a drop of nearly 37 percentage points. At maximum effort, the error rate fell by 18 percentage points to 4.3%, nearly matching the rates for GPT-6 Astra and GPT-6.1 Sol.

On coding, OpenAI says GPT-6.1 Terra marks a substantial improvement over GPT-5.6 Terra for software engineering. On Terminal-Bench 4.0, which tests how well coding agents solve command-line problems, GPT-6.1 Terra at maximum reasoning effort solves 50.3% of tasks, more than double GPT-5.6 Terra's published score of 21.5%. OpenAI adds that GPT-6.1 Sol and GPT-6 Astra remain the better choice for the most complex coding tasks. On Terminal-Bench-Science 0.1, which tests how well agents use code for scientific research tasks, GPT-6.1 Terra at max reasoning surpasses GPT-6.1 Sol at lower cost and comes within five points of Opus 5.5 at less than one-fifth the estimated cost.

Safety and alignment

OpenAI says GPT-6.1 Terra meets the Critical threshold in cybersecurity under its Preparedness Framework and maintains the protections developed for GPT-6 Astra. OpenAI also says Terra brings the alignment improvements of the GPT-6 series: it is more likely to stay within scope and honestly disclose limitations and potential blockers to the user compared with GPT-5.6 Terra. In prompt-injection tests, OpenAI says Terra showed strong resistance to malicious instructions designed to divert it from the user's task. More detail is available in the GPT-6.1 Terra system card at deploymentsafety.openai.com/gpt-6-1-terra.

The Hacker News reaction: naming confusion and the middle-child problem

The most repeated criticism in the Hacker News thread, which had 629 points and 424 comments at scrape time, was that OpenAI's model lineup has become hard to reason about. Commenters traced a lineage from Luna as small, Terra as medium, Sol as big, and Astra as bigger, then noted that Sol-6.1 sometimes outperformed Astra while Terra was discontinued, only for Terra to return as a 6.1 release.

“I'm very confused about their model market position/naming schema,” wrote commenter Someone1234. “Did Microsoft's chief branding officer move over to OpenAI?”

A related criticism was that Terra occupies an awkward middle tier. Commenter minimaxir called Terra “a weird middle child between the small-model-that-got-substantially-better (Luna) and the large-model-that-got-significiantly-cheaper (Sol).” Commenter Silagi argued that most tasks are better served by going up a tier and down a reasoning level, or down a tier and up a reasoning level: “Virtually every task I give to either is better served by going up a tier and down a reasoning level, or down a tier and up a reasoning level.”

Several commenters said they could not identify a clear niche for Terra where a cheaper model with more reasoning, or a more expensive model at the same cost, did not produce equal or better results. Others framed the release as a response to Anthropic's Sonnet 5.5, which commenters described as having its own middle-child problem between Opus 5.5 and Haiku 5.5.

Cost-per-task claims meet benchmark skepticism

Commenters were split on whether OpenAI's Pareto-frontier charts translate into practical model choice. Some found the cost-per-task framing useful. Commenter Archit3ch wrote, “I rather enjoy OpenAI's focus on cost-per-task.” But others argued that performance-versus-price charts embed subjective choices.

“These ‘Pareto frontier’ performance vs price graphs have become very popular and I find them mostly uninformative. Pricing is subjective. You chose the position on the X axis, and which measures to use on the Y Axis,” wrote commenter causal.

Commenter simianwords said the benchmarks are already saturated and that the curves for Terra stick to the axes. Commenter ralusek pointed to Artificial Analysis and said, “According to this, Sol 6.1 x-high is both cheaper and smarter than Terra 6.1 max.” Commenter XCSme reported a different result from hands-on testing: “in my tests, on max effort, it's worse than 6.1 sol and 2x more expensive,” while “Used on medium seems to be the best use-case and cheaper than 6.1 Sol.”

Where commenters saw a fit

Some commenters found a clear niche for Terra despite the positioning confusion. Commenter mehdizare said the tier is what they would point opencode at for PR reviews because “The top models are too slow and expensive to run on every diff.” Commenter bob1029, who said their custom enterprise agent runs on GPT-5.6 Terra, called the factual-error-rate improvement “staggering” and said they looked forward to updating the configuration.

Commenter chriddyp reported that in data-analytics benchmarks, GPT-6.1 Terra “joins the Pareto Frontier in between GPT-6 Luna and GPT-6.1 Sol. So, exactly as advertised!” chriddyp also said Terra was very fast and wrote the least output in its responses among the models tested. Commenter thefourthchime reported PacBench results in which GPT-6.1 Terra scored 84, one point over Grok 4.7 at 83, while completing the task in 5.4 minutes for $0.19 versus Grok's 36.5 minutes and $1.69.

Not all hands-on reports were positive. Commenter wellthatsinsa said the model “used lots of unnecessary skills and read through way more memory files needed. It literally burned through 2 subs.” Commenter agluszak said Terra suffers from the same problem as Luna: “it will stop at an earliest excuse and no amount of prompting not to stop until X is done will convince it otherwise.” Commenter notenchange said Sol 6.1 kept refusing GPU-unlocking work over “cyber concerns” and planned to try Terra while hoping it complained less about what the commenter called non-existent cybersecurity risks.

Strategic context: a cheaper tier, not a frontier answer

The release reads as a cost-efficiency play rather than an attempt to beat the top of the market. OpenAI's own announcement says GPT-6.1 Sol and GPT-6 Astra remain better choices for the most complex coding tasks. Commenter enraged_camel summarized the strategic tension: “People keep asking OpenAI for a mode that's as good as Opus 5.5, and they release this instead. Interesting strategy. Let's see how it works out for them.” Commenter guybedo put it more bluntly: “OpenAi needs a model to compete with Opus 5.5, and that's not Terra 6.1.”

Commenters also questioned the broader trend of model proliferation. Commenter xnx asked why Anthropic and OpenAI split their lineups into four versions while Google uses two, and commenter mrheosuper counted “like 5 models and 5 thinking levels, that is 25 different combinations.” Commenter ulfw asked whether people still care about updated LLM versions at all: “It's all a commodity now.”

The practical takeaway from the thread is that GPT-6.1 Terra is most compelling for high-volume, medium-complexity work where latency and cost per task matter more than maximum capability. Its strongest advocates pointed to PR review, enterprise agents, data analytics, and other repeated workflows. Its critics argued that the model's position between Luna and Sol is still hard to justify without task-specific testing, and that OpenAI's rapidly changing model names make long-lived production choices harder to maintain.