GPT-6.1 Terra halves OpenAI's mid-tier input price

OpenAI released GPT-6.1 Terra on 26 October, a fast, efficient model it pitches at computer use, visual reasoning and clearer writing [1][3]. In the API it costs $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens [1][2]. In ChatGPT it is rolling out to Plus, Pro, Business, Enterprise and Edu plans in the Work and Codex experiences, with each plan’s overall usage limits unchanged [3]. For a solo operator or a small team, the questions are simple: is it cheaper than what you use now, is it good enough for your work, and does it change anything in the ChatGPT plan you already pay for.

What was announced

OpenAI describes GPT-6.1 Terra as a fast, efficient model that combines detailed visual reasoning and precise computer use with clearer, more natural writing [1]. The company says it is aimed at the communication and computer-based tasks that fill a workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos [1]. Its example of high-volume work is reading incoming invoices and entering their details into accounting software [1].

In the API the model is called gpt-6.1-terra [1][2]. The model page lists a 1,050,000-token context window, 128,000 maximum output tokens and a knowledge cutoff of 1 June 2026 [2]. It takes text and images as input and returns text; audio is not supported [2]. Reasoning effort can be set to none, low, medium, high, xhigh or max, with medium as the default, and tool calling requires the Responses API [2].

The standard price is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens [1][2]. Writing to the prompt cache costs $1.25 per million tokens for 30-minute retention and $2.50 for 12-hour retention [2]. Prompts with more than 272K input tokens are billed at twice the input and cache rates and 1.5 times the output rate for the whole request [2]. Batch and Flex prices are 50% lower than Standard, Fast mode costs twice Standard, and regional processing adds a 10% premium where available [2]. On the pricing page, Batch Terra comes to $0.50 input and $2.50 output per million tokens [4].

For comparison, GPT-5.6 Terra, the model in the same slot before, lists at $2.00 per million input tokens, $0.20 cached and $12.00 output [5]. On the current pricing page, GPT-6.1 Sol lists at $2.00 input and $10.00 output, GPT-6 Astra at $10.00 and $50.00, and GPT-6 Luna at $0.10 and $0.50 [4]. OpenAI’s announcement says Terra’s standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol [1]. On the listed numbers that holds exactly for input; Terra’s output price is less than half of GPT-5.6 Terra’s [2][5].

In ChatGPT, the release notes say Terra is rolling out in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu plans, and that you select it from the model list as it reaches your account [3]. Standard and Fast modes are supported where available [3]. Terra does not support UAE inference residency, a restriction based on where inference runs rather than where you are [3]. The announcement’s own list of plans names Plus, Pro, Enterprise and Edu, while the release notes add Business [1][3].

What changed, and for whom

The first change is the price per token for mid-tier API work. If you run a steady job through GPT-5.6 Terra, such as classifying support emails or extracting fields from documents, the listed input price halves and the output price falls from $12.00 to $5.00 per million tokens [2][5]. If you moved up to GPT-6.1 Sol for quality, Terra lists at half Sol’s input and output prices [2][4]. Whether that saves money depends on how many tokens Terra spends on your task. OpenAI says that on its Agents’ Last Exam benchmark at medium effort, Terra scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost, and performs on par with GPT-6 Astra at one-tenth its price per token [1]. Those are the company’s own measurements, on its own benchmark.

The second change is in ChatGPT. Your plan’s overall usage limits do not change [3]. What changes is how far they go. OpenAI says the same allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with GPT-6.1 Sol or GPT-6 Astra [1]. The release notes add a detail that may surprise you: some Plus and Pro users may see GPT-6.1 Terra at Medium instead of GPT-6.1 Sol at Light in the default Power presets in Work [3]. If your answers feel different this week, check which model the preset picked.

The third change is quality claims, and these need the most care. OpenAI says that on its internal benchmark of error-inducing prompts, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra, and at maximum effort fell to 4.3% [1]. It says Terra solves 50.3% of Terminal-Bench 4.0 tasks at maximum effort, against GPT-5.6 Terra’s published 21.5% [1]. It also says that for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain the better choice [1]. On safety, OpenAI says Terra meets the Critical threshold in cybersecurity under its Preparedness Framework and keeps the protections developed for GPT-6 Astra [1].

An outside view helps place it. Simon Willison notes that Terra sits in the same position as Sonnet in Anthropic’s line-up, neither the cheapest nor the most expensive, which makes it harder to decide when to use it [6]. He also points out that computer use is a good fit for a cheaper model, because it tends to be automated, repetitive work [6]. He quotes OpenAI’s Ted Sanders, writing on Hacker News, describing Terra as probably better than Astra and Sol at technical communication, similar on vision and computer use but much faster, and a bit worse on most other things [6].

Who it is for — and not

Terra suits work where volume matters more than depth. OpenAI’s own examples are reading invoices into accounting software, drafting proposals, interpreting technical drawings and filling out forms [1]. If you run an automation that calls a model thousands of times a month, the per-token price is where Terra helps most. If you use ChatGPT for drafting, summarising and screen-based tasks and keep hitting your limits, Terra is the model to try first [1][3].

It is not the model for your hardest coding work; OpenAI says Sol and Astra remain the better choice there [1]. It is not a model for audio, which the API page lists as not supported [2]. It is not available where you need UAE inference residency [3]. And it is not a reason to switch providers on its own: the benchmark numbers are OpenAI’s, and the only test that counts is your own task.

Two pricing details are easy to miss. Long prompts cost more: above 272K input tokens the whole request is billed at twice the input rate and 1.5 times the output rate [2]. And the reasoning effort is a setting with six levels, from none to max, so choose it on purpose rather than leaving every job at the default [2].

This page will be updated if OpenAI changes Terra’s price, adds plans or regions, or publishes independent benchmark results, or announces a retirement date for GPT-5.6 Terra.

∴ Terra is now OpenAI’s cheaper middle model for API work; in ChatGPT it stretches the same usage limits further.