GPT-6.1 Terra: $1/$5 per million tokens, half GPT-6.1 Sol's price, in ChatGPT Work, Codex and the API

GPT-6.1 Terra is OpenAI's new mid-priced model, released on October 26, 2026. In the API it costs $1 per million input tokens and $5 per million output tokens, exactly half of GPT-6.1 Sol's price, and it's available in ChatGPT Work and Codex on paid plans. OpenAI pitches it as a fast, efficient model for computer use, image understanding and writing. It's the first Terra model since GPT-5.6 Terra in July: the GPT-6 generation launched without one, as our GPT-6 model comparison shows.

This story is based on OpenAI's launch post, its API pricing page, its GPT-6.1 Terra, GPT-6.1 Sol and GPT-5.6 Terra model pages, its API and ChatGPT/Codex changelogs, the GPT-6.1 Terra system card addendum and OpenAI's developer-forum announcement, plus reporting by 9to5Mac, checked October 27, 2026.

The short version

• Price: $1 input, $0.05 cached input and $5 output per million tokens, half of GPT-6.1 Sol's $2/$0.10/$10.
• Versus GPT-5.6 Terra: OpenAI says Terra's prices are 50% lower than GPT-5.6 Terra's. On the published rates, input is 50% lower and output 58% lower ($5 against $12).
• Where: ChatGPT Work, Codex and the API (gpt-6.1-terra). Not in the ChatGPT Chat tab.
• Specs: 1,050,000-token context window, 128,000 maximum output tokens, a June 1, 2026 knowledge cutoff, text and image input.
• Benchmarks: all vendor-reported. OpenAI still recommends Sol or Astra for the hardest coding and research.

How much does GPT-6.1 Terra cost?

OpenAI's pricing page lists GPT-6.1 Terra second in its four-model lineup. 9to5Mac describes the order, from cheapest and least capable to most expensive and most capable, as Luna, Terra, Sol and Astra.

Last verified: October 27, 2026

Other tiers follow OpenAI's usual multipliers. Batch and Flex cost $0.50 input and $2.50 output, and Fast mode costs $2 input and $10 output, the same as Sol at standard speed. OpenAI lists no Ultrafast price for Terra. Prompts over 272,000 input tokens cost 2x on input and 1.5x on output for the whole request, and regional processing adds 10%, according to the model page.

The "50% less" claim needs one qualification. GPT-5.6 Terra's model page lists $2 input, $0.20 cached input and $12 output, the rates since OpenAI cut them by 20% on July 30, 2026. Terra's new input price is half that, its cached input a quarter, and its output price about 42% of it. Why prices across OpenAI's lineup keep falling is the subject of our analysis of the 2026 price cuts.

Where can I use GPT-6.1 Terra?

OpenAI's ChatGPT and Codex changelog says, for October 26: "GPT-6.1 Terra is now available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans." It adds that the model "isn't available in Chat" and doesn't support UAE inference residency, and that Enterprise access depends on workspace availability and admin controls. OpenAI's launch post and developer-forum announcement list Plus, Pro, Enterprise and Edu, without Business. Free and Go plans aren't on either list. Our ChatGPT plans comparison covers what each plan costs.

In Codex, Terra uses your allowance more slowly than Sol. OpenAI's Codex pricing page charges 25 credits per million input tokens and 125 per million output tokens for Terra, against 50 and 250 for GPT-6.1 Sol and 50 and 300 for GPT-5.6 Terra. It estimates 35 to 285 local messages per five hours on Plus and standard Business seats for Terra, against 15 to 160 for Sol.

In the API, Terra works with the Responses and Chat Completions endpoints and supports web search, file search, computer use, MCP and the other hosted tools. Its rate limits match GPT-6.1 Sol's: 1 million tokens per minute on the Build tier up to 40 million on Grow.

GPT-6.1 Sol vs Terra: which should you use?

The two models share a context window and maximum output, and differ mainly in price, reasoning range and what OpenAI says each is for.

Last verified: October 27, 2026

OpenAI draws the line itself. Its forum announcement recommends GPT-6.1 Sol or GPT-6 Astra for in-depth research and complex coding, and says Terra's allowance in Work and Codex goes "far" further than Sol's or Astra's. For a full comparison of Sol, Astra and Luna, see our GPT-6 model guide; for the launch of Sol itself, see our GPT-6.1 Sol coverage.

What do OpenAI's benchmarks say?

Every result below is vendor-reported: OpenAI ran it on its own models and published it in its launch post on October 26, 2026. The configurations are only as detailed as that post.

• Errors on hard prompts: on an internal OpenAI test of prompts designed to induce errors, OpenAI says the error rate at low reasoning effort fell from 53.6% for GPT-5.6 Terra to 16.7% for GPT-6.1 Terra, and to 4.3% at maximum effort, close to GPT-6 Astra and GPT-6.1 Sol. The test isn't public.
• Computer use: on Agents' Last Exam, a benchmark of professional tasks done with workplace software, OpenAI says Terra at medium reasoning effort scores 15 percentage points higher than GPT-5.6 Terra at about one-third the cost. It also says Terra performs on par with GPT-6 Astra at a tenth of Astra's price per token, and finishes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, without giving the effort settings for those two claims.
• Coding: OpenAI says Terra solves 50.3% of Terminal-Bench 4.0 tasks at maximum effort, against GPT-5.6 Terra's published 21.5%. It adds that Sol and Astra "remain the better choice" for the most complex coding.

OpenAI's system card addendum, published the same day, adds some figures 9to5AI could read directly. On HealthBench Professional, a length-adjusted health-conversation evaluation, GPT-6.1 Terra scored 62.7 against 57.7 for GPT-5.6 Terra and 64.2 for GPT-6.1 Sol. OpenAI rates Terra "Critical" for cybersecurity and "High" for biological and chemical capability under its Preparedness Framework, and says it ships with the same safeguards as GPT-6.1 Sol and GPT-6 Astra. No independent evaluator's results were read for this story.

What's confirmed, reported and uncertain

• Confirmed (OpenAI pages): the October 26 release; all API prices and tiers; the specs and rate limits; availability in Work, Codex and the API but not Chat; the Codex credit rates; the system card figures and Preparedness ratings.
• Vendor-reported (OpenAI's launch post, not independently checked): the 53.6%, 16.7% and 4.3% error rates, the Terminal-Bench 4.0 scores and the Agents' Last Exam gain.
• Uncertain: whether ChatGPT Business has Terra (OpenAI's changelog includes it; its launch post and forum announcement don't); whether GPT-5.6 Terra will be retired (no deprecation is listed); and when, if ever, Terra reaches the Chat tab. For what OpenAI has said about its next big model, see our GPT-7 release date explainer.