Model comparisons

GPT-6.1 Sol vs GPT-6.1 Terra: Which OpenAI Model Should You Use?

GPT-6.1 Terra at $1/$5 per million tokens costs half of GPT-6.1 Sol at $2/$10, uses 25 instead of 50 Codex input credits and roughly doubles the Plus five-hour estimate. Compare prices, ChatGPT availability and which work still needs Sol.

UpdatedOctober 26, 2026

In short

Use GPT-6.1 Terra for routine coding edits, screen and image work, computer use and drafting, and keep GPT-6.1 Sol for complex coding and multi-step professional work where a failed attempt costs more than the price difference. Terra, released on October 26, 2026, costs half of Sol at every published rate: $1 per million input tokens and $5 per million output in the API, against Sol's $2 and $10, and 25 Codex credits per million input tokens against 50. On ChatGPT Plus that roughly doubles the five-hour estimate, from 15–160 local messages on Sol to 35–285 on Terra. Both run in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu, and neither is available in Chat.

On this page5

Terra costs half of GPT-6.1 Sol at every API rate

Rate, per million tokens

GPT-6.1 Sol

GPT-6.1 Terra

Standard input / output

$2 / $10

$1 / $5

Cached input

$0.10

$0.05

Cache write, 30 minutes / 12 hours

$2.50 / $5

$1.25 / $2.50

Batch and Flex input / output

$1 / $5

$0.50 / $2.50

Fast mode input / output

$4 / $20

$2 / $10

Ultrafast input / output

$12 / $60

Not offered

Codex credits: input / cached / output

50 / 2.5 / 250

25 / 1.25 / 125

Read the table as one ratio: every Terra price is half the Sol price on the same row, so switching a workload halves its token bill at the same token counts. Both models bill prompts above 272,000 input tokens at their long-context rates, twice the input and cache prices and 1.5 times the output price for the whole request. Ultrafast is the one row where Terra has no price, because OpenAI offers it only for GPT-6.1 Sol and GPT-6 Astra.

On Plus, Terra stretches the five-hour allowance about twice as far

OpenAI's Codex pricing page gives Plus and Standard Business seats these estimates of local messages per five hours: 15–160 on GPT-6.1 Sol and 35–285 on GPT-6.1 Terra. OpenAI says overall Work and Codex plan limits did not change with Terra's release, so the gain comes from Terra drawing less of the same allowance per message, not from a bigger allowance. Pro plans currently have no five-hour limit, and weekly limits may apply on any plan.

An illustrative agent turn shows the same halving in money and credits. Take 200,000 input tokens, of which 160,000 are read from cache, and 10,000 output tokens:

Illustrative turn

GPT-6.1 Sol

GPT-6.1 Terra

API cost

$0.196

$0.098

Codex credits

4.9

2.45

2,000 such turns a month, API

$392

$196

The workload is ours, not OpenAI's. The saving only holds if Terra finishes the task in a similar number of attempts; a task that needs a second Terra run costs about what one Sol run does.

OpenAI pitches Terra at screens and images, Sol at complex coding

OpenAI describes GPT-6.1 Terra as a fast, efficient model for visual reasoning, precise computer use and clear, natural writing, with examples such as matching product photos to catalog entries, interpreting technical drawings, acting on application screens and drafting or revising text. It released GPT-6.1 Sol for complex coding and professional work at a lower cost than GPT-6 Astra. Neither model page lists a benchmark score that compares the two, so test Terra on your own tasks before moving work that Sol handles well today.

The technical limits are close. Both take up to 1,050,000 tokens of context and return up to 128,000. Terra accepts text and image input, and its reasoning effort runs from none to max with medium as the default, while Sol has no none setting. Terra's knowledge cutoff is June 1, 2026, a month after Sol's April 30. In the API, GPT-6.1 Sol also supports multi-agent work in the Responses API as a beta.

A practical split, which is our judgment rather than OpenAI's guidance:

  • Terra first: screenshot-driven UI fixes, browser and desktop actions, reading charts or drawings, routine edits in a familiar repository, and rewriting documentation or reports.
  • Sol first: multi-file changes in unfamiliar code, long agent runs where a wrong turn is expensive, work that needs Ultrafast, and tasks where Terra's first attempts fail your checks.
  • Neither: high-volume extraction and classification usually fits GPT-6 Luna, at a tenth of Terra's API price.

Both run in Work and Codex on paid plans, not in Chat

In ChatGPT, GPT-6.1 Terra is available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu. You choose it under Advanced in the model list in the ChatGPT desktop app or in ChatGPT Work on the web, and Work also supports it on mobile. Enterprise access depends on workspace availability and administrator controls, and Terra does not support UAE inference residency. GPT-6.1 Sol sits on the same plans and surfaces, and on Enterprise and Edu it stays off until an administrator enables it.

GPT-6.1 Sol and GPT-6.1 Terra are not available in Chat, which runs GPT-6 Sol on paid plans. In the API the model IDs are gpt-6.1-sol and gpt-6.1-terra, billed to an OpenAI Platform account separately from any ChatGPT plan.

GPT-6.1 Terra is not the Terra on ChatGPT Go

ChatGPT Go ($8 a month) lists limited GPT-5.6 Terra in Work and Codex on desktop. That is the previous Terra, and in Codex it uses 50 credits per million input tokens, 5 cached and 300 output, two to four times GPT-6.1 Terra's rates. GPT-6.1 Terra is not included on Free or Go; it starts at Plus for $20 a month.

Official sources

Editorial guidance grounded in official product sources.

FAQ

Common questions

Is GPT-6.1 Terra cheaper than GPT-6.1 Sol?

Yes, by half at every published rate. In the API GPT-6.1 Terra costs $1 per million input tokens, $0.05 cached and $5 output, against $2, $0.10 and $10 for GPT-6.1 Sol. In Codex it uses 25 credits per million input tokens, 1.25 cached and 125 output, against 50, 2.5 and 250.

How many GPT-6.1 Terra messages do I get on ChatGPT Plus?

OpenAI estimates 35–285 local messages per five hours in ChatGPT Work and Codex on Plus and Standard Business seats, against 15–160 for GPT-6.1 Sol. These are estimates, not fixed limits; weekly limits may also apply, and Pro plans currently have no five-hour limit.

Which ChatGPT surfaces offer GPT-6.1 Terra?

No. GPT-6.1 Terra is available in ChatGPT Work and Codex on Plus, Pro, Business, Enterprise and Edu, and is not available in Chat. You pick it under Advanced in the model list in the ChatGPT desktop app or in ChatGPT Work on the web; Work also supports it on mobile.

Is the Terra on ChatGPT Go the same as GPT-6.1 Terra?

No. Go lists limited GPT-5.6 Terra in Work and Codex on desktop, an older model that uses 50 Codex credits per million input tokens and 300 per million output. GPT-6.1 Terra is not included on Free or Go.

Does GPT-6.1 Terra support Ultrafast?

No. OpenAI's API pricing lists Ultrafast only for GPT-6 Astra and GPT-6.1 Sol, and in ChatGPT only Pro 500 and eligible Enterprise and Edu plans get Ultrafast. Fast mode is available for Terra at twice its Standard price.

Next steps

Open both sides of the distinction