AI news · 2 min read
Copilot adds GPT-6.1 Terra for everyday coding, with usage costs to check
GPT-6.1 Terra is rolling out across paid Copilot plans. Compare token rates, access policy and a small real task before switching your default.
Announcement date:
Affiliate disclosure: Some links may earn CreatorStack a commission at no extra cost to you.
At a glance
GitHub’s October 26 release makes GPT-6.1 Terra generally available through a gradual rollout to paid Copilot plans. At up to 272K input tokens, listed rates are $1 per million input and $5 per million output tokens, with separate cache rates. Check your model policy and compare a small task before making it the default.
A broken signup check does not always need the most expensive coding model. It needs a change that accepts the right input, rejects the wrong input and leaves the rest of the form alone. GitHub’s October 26 GPT-6.1 Terra release adds another option for that kind of bounded work in Copilot. Release details.
GitHub positions it for everyday coding and terminal tasks. Its early performance observations are vendor testing, not evidence that Terra will fix your project better than your current model.
Availability and cost are separate checks
The announcement covers Pro, Pro+, Max, Business and Enterprise, across Copilot’s listed IDE, CLI, cloud, web and mobile surfaces. Rollout is gradual. Team model policies can still affect access, and the supported-model table does not yet supply Terra’s minimum IDE versions.
The Copilot pricing table lists these USD rates per million tokens:
| Input-context tier | Input | Cached input | Cache write | Output |
|---|---|---|---|---|
| Up to 272K | $1.00 | $0.05 | $1.25 | $5.00 |
| Above 272K | $2.00 | $0.10 | $2.50 | $7.50 |
These are usage rates, not a monthly subscription price or a fixed charge for a completed fix. GitHub converts usage into AI credits and applies plan allowances. A conversation can accumulate context, retries and tool results, so a lower token rate does not tell you the final cost of a task by itself.
Give the comparison a finish line
Try a small, reversible issue whose expected behavior you can state before asking for code. For an illustrative form bug, use one valid submission and one invalid submission that currently slips through. Keep that requirement the same when comparing models.
Review the changed behavior and the resulting usage together. A quick answer that needs several corrections may be a worse fit than a slower first answer that passes the checks. This is a proposed comparison, not a benchmark we ran.
Check the picker and your team’s policy first. Then compare one task before changing a default used for larger jobs. Our model-policy API coverage explains the access side; the AI coding evaluation workflow helps define a result worth keeping.
Common questions
Questions people ask.
- Is this an announcement of a separate OpenAI API rollout?
- This article covers GitHub Copilot. Access and billing on a standalone model API are separate questions.
- Does generally available mean it is already in every account?
- GitHub says rollout is gradual. Business and Enterprise model policies can also affect access; the checked minimum-version table still lists Terra values as TBD.
Sources & freshness
This article was refreshed on 27 Oct 2026 using the sources below. A source check verifies published information, not output quality or checkout availability in every region.
- github.blog/changelog/2026-10-26-gpt-6-1-terra-in-github-copilot/
- docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing
- docs.github.com/en/copilot/reference/ai-models/supported-models
What changed: Original coverage of Terra’s paid-plan Copilot rollout, token tiers and a practical comparison against an existing coding workflow.
Found a correction? Let us know.