GPT-6.1 Terra: OpenAI's Faster, Cheaper Work Model

Updated: October 2026 · 6 min read · By the CreatorSprout team

Launch coverage based on OpenAI's announcement and independent launch reporting — we have not tested this model hands-on. Vendor figures and independent measurements are kept separate below, because they tell different stories. Pricing verified October 2026; confirm on official pages before you commit. No affiliate links in this article.

The interesting part of this launch is not a new capability. It is the price-to-speed ratio: OpenAI is positioning Terra as the model you run all day, every day, instead of reserving the expensive models for everything.

What GPT-6.1 Terra is

GPT-6.1 Terra is a reasoning model in OpenAI's four-tier lineup, which runs from cheapest to most capable as Luna, Terra, Sol and Astra. It accepts text and image input and produces text, with a 1,050,000-token context window, up to 128,000 output tokens, and a June 1, 2026 knowledge cutoff. Reasoning effort runs from none to max through the Responses API, with medium as the default — and at the none setting, temperature and top_p become adjustable, which was not possible on GPT-6.1 Sol.

OpenAI's own examples of its intended work are practical rather than spectacular: drafting proposals, interpreting technical drawings, filling out forms, editing videos, matching product photos to catalog entries, and reading incoming invoices and typing the details into accounting software. In short, computer use — an AI that looks at a screen and clicks, types and moves between apps — plus clearer, more natural writing than its predecessor.

For developers, tool calling requires the Responses API (not just a model-string swap in Chat Completions), with supported tools including web search, file search, image generation, code interpreter, hosted shell, computer use and MCP. Fine-tuning is not supported, so teams steer it through prompts, tools and workflow design instead.

What OpenAI claims

Every number in this section comes from OpenAI. No independent benchmark results were available at launch, and that matters — treat these as the company's best case until outside evaluators publish their own runs.

Agents' Last Exam (computer-use work): at medium effort, Terra scores 15 percentage points above GPT-5.6 Terra at roughly a third of the cost per task, and matches GPT-6 Astra at one-tenth the price per token.
Speed: OpenAI says Terra finishes tasks more than three times faster on average than both GPT-6 Astra and GPT-6.1 Sol, based on its own timing runs.
Terminal-Bench 4.0 (agentic coding): 50.3% of tasks solved at max effort, versus 21.5% published for GPT-5.6 Terra.
Fewer factual errors: on OpenAI's internal set of prompts designed to trigger errors, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7%, and to 4.3% at max effort — nearly matching Astra and Sol.
Usage allowance: in ChatGPT Work and Codex, OpenAI says the same plan allowance covers more tasks with Terra than with GPT-5.6 Terra, and substantially more than with Sol or Astra.

One early outside reaction is worth noting, with the same caveat: Artificial Analysis posted that Terra is "the first Terra-class model since GPT-5.6, and is more intelligent for less than a third of the cost." That is an evaluator's first impression, not a full independent benchmark — the detailed scores are still to come.

Pricing: cheap — with a long-prompt catch

Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. OpenAI describes that as 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. Cache writes cost $1.25 per million tokens for 30-minute retention and $2.50 for 12-hour retention.

The catch sits in the long-context surcharge, and it is easy to miss in a cost model that only stores one input rate: prompts over 272,000 input tokens are billed at twice the input and cache rate and 1.5 times the output rate for the whole request — $2 input and $7.50 output per million tokens. A 1.05M-token window is real, but the headline rate does not apply across all of it.

Two more pricing dials: Batch and Flex processing cut standard prices by 50%, while Fast mode doubles them. Regional processing, where available, adds a further premium. As with every model in this series, price per token is not price per outcome — measure cost per completed task on your own workload, not per million tokens.

For ChatGPT users there is no separate Terra price: it is included in the Plus, Pro, Business, Enterprise and Edu plans' Work and Codex access, with availability depending on plan, workspace settings and rollout progress, so some users may see it appear over days rather than all at once.

How to access it

ChatGPT Work or Codex: pick it from the model list in ChatGPT Work or the desktop app, or run codex -m gpt-6.1-terra in the Codex CLI (version 0.159.0 or later).
Developers: use the API model id gpt-6.1-terra through the Responses API for tool use.
Also on AWS Bedrock: Amazon made it available on Bedrock the same day, for teams already running there.
Start with high-volume computer work. That is where the price and speed story is strongest — forms, invoices, drafting, screen-reading tasks — not the hardest coding.
For deep research or the most complex coding, OpenAI itself still points to GPT-6.1 Sol or GPT-6 Astra. Terra is the everyday model, not the flagship.

The honest caveats

We haven't tested it. Every performance number in this article comes from OpenAI or launch reporting, not from our own runs.
No independent benchmarks yet. The speed and accuracy claims are OpenAI's own timing runs and internal tests. Wait for independent evaluations before treating them as settled.
Not in regular Chat, no free tier. If you only use free ChatGPT, this launch does not change anything for you.
The headline price has a threshold. Past 272,000 input tokens, the whole request reprices upward. Long-document workflows need the $2/$7.50 math, not the $1/$5 headline.
No fine-tuning. Teams that rely on fine-tuned models will need to steer Terra with prompts and tools instead.
OpenAI classifies Terra at the Critical cybersecurity capability threshold under its Preparedness Framework. That is the company's own capability assessment, and it is one more reason to treat computer-use access — what the model is allowed to click and change — as something to permission carefully, not hand over blindly.

Verdict

GPT-6.1 Terra is the most practical model launch of the week — not because it tops any chart (on OpenAI's own framing, Sol and Astra still own the hardest work), but because it makes all-day computer-use agents affordable: a mid-tier model at $1/$5, over three times faster than the flagships on OpenAI's timing, with a million-token window for long, screen-heavy jobs. The honest move: if you already pay for ChatGPT Work, Codex or the API, try it first for your repetitive computer tasks and measure cost per finished job — and wait for independent benchmarks before believing the speed and accuracy claims in full.

FAQ

What is GPT-6.1 Terra?

A mid-tier AI model launched by OpenAI on October 26, 2026, built for visual reasoning, computer use and clearer writing. It replaces GPT-5.6 Terra in OpenAI's lineup and is available in ChatGPT Work and Codex for paid plans, and through the API as gpt-6.1-terra.

How much does GPT-6.1 Terra cost?

Through the API, $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens at standard rates. Requests over 272,000 input tokens are billed at higher rates, and Batch and Flex processing cost 50% less while Fast mode costs twice as much.

Is GPT-6.1 Terra available in regular ChatGPT?

No. OpenAI lists it for ChatGPT Work and Codex for Plus, Pro, Business, Enterprise and Edu users. It is not in regular Chat, and no free tier is listed.

How good is GPT-6.1 Terra?

On OpenAI's own figures, it beats GPT-5.6 Terra clearly on computer-use and coding tests and matches GPT-6 Astra on one agent benchmark at a tenth of the price per token, while finishing tasks over three times faster than Astra and GPT-6.1 Sol. Independent evaluations were not yet available at launch, so these remain vendor claims.

Should I switch to GPT-6.1 Terra?

If your work is high-volume computer tasks — forms, invoices, drafting, reading screens — and you already pay for ChatGPT Work, Codex or the API, it is the obvious model to try first. For the hardest coding or research, OpenAI itself still points users to GPT-6.1 Sol or GPT-6 Astra.