← All posts
Guide

GitHub Copilot drops Claude Opus 4.8, Opus 5 and GPT-5.6 Terra on 16 November: what to switch to and what it costs

Three Copilot models go on 16 November and six more on 19 October. The suggested replacements, their token prices, and what admins need to check.

5 min read6 sources

GitHub is removing three more models from Copilot next month, and six others go this coming Monday. If your team pins a model in Copilot settings, scripts or agent configs, or you switched off GitHub’s default model enablement, there is something to check before each date.

What GitHub announced on 15 October

GitHub’s changelog says three models will be deprecated across all Copilot experiences on 16 November 2026: Copilot Chat, inline edits, ask and agent modes, and code completions.

Model leaving on 16 Nov GitHub’s suggested alternative
Claude Opus 4.8 Claude Opus 5.5
Claude Opus 5 Claude Opus 5.5
GPT-5.6 Terra GPT-6.1 Sol or GPT-6 Luna

There is one exception. Individual subscribers on annual plans keep Claude Opus 4.8, which GitHub says is so they still have an Opus option.

Sooner: six models go on 19 October

An earlier changelog post set a separate batch for Monday, 19 October 2026:

Model leaving on 19 Oct Suggested alternative
Gemini 3.7 Flash Gemini 3.8 Flash
GPT-5.5 GPT-5.6 Sol
GPT-5.4 GPT-5.6 Sol
GPT-5.4 mini GPT-5.6 Luna
GPT-5 mini GPT-5.6 Luna
Grok 4.5 Grok 4.6

If anyone on your team still picks GPT-5.4 mini or GPT-5 mini as a cheap default, this is the change that will reach them first.

What the replacements cost

Copilot now bills by usage. GitHub’s models and pricing page says each interaction is priced on input, output and cached tokens and converted to GitHub AI Credits, at 1 credit = US$0.01. Individual plans include an allowance; Business and Enterprise allowances are per user but pooled across the billing entity. Usage beyond that is billed at the per-token rates.

Here are the listed rates, in US$ per million tokens, for the models leaving on 16 November and their replacements:

Model Input Cached input Cache write Output
Claude Opus 4.8 5.00 0.50 6.25 25.00
Claude Opus 5 5.00 0.50 6.25 25.00
Claude Opus 5.5 4.00 0.20 5.00 20.00
GPT-5.6 Terra (≤272K input) 2.00 0.20 2.50 12.00
GPT-6.1 Sol (≤272K input) 2.00 0.10 2.50 10.00
GPT-6 Luna (≤272K input) 0.10 0.01 0.125 0.50

Above 272K input tokens the OpenAI rates step up: Terra to $4 / $18, Sol to $4 / $15 and Luna to $0.20 / $0.75 (input / output).

Three things stand out:

  • The Opus move is cheaper on every line. At GitHub’s prices, Opus 5.5 costs less than Opus 4.8 and Opus 5 for input, output and cache writes, and less than half for cached input. Moving the Opus workloads should not raise your bill, all else equal.
  • Sol is the closer price match for Terra. Input and cache-write prices are the same, output is $2 cheaper, and cached input is half the price.
  • Luna is a different tier. Its input price is one-twentieth of Terra’s. That makes it attractive for high-volume chat and simple edits, but it is not a drop-in quality match by price alone. Try it on your own repos before you move agent-mode work to it.

GitHub also notes that code completions and next edit suggestions are not billed in AI credits.

Annual Pro and Pro+ plans. The legacy multiplier page says request-based multipliers now apply only to Copilot Pro and Pro+ subscribers on an existing annual plan, who stayed on that model after 1 June 2026. It lists Claude Opus 4.8 at a multiplier of 27 and does not list Opus 5 or 5.5. That matches GitHub’s reason for letting these subscribers keep 4.8.

What admins need to check

From the 15 October changelog:

  1. Default model enablement. On Copilot Business and Enterprise, the suggested alternatives are switched on automatically, unless an admin has turned off the global default or explicitly disabled that model.
  2. If you turned the global default off, nothing is enabled for you. Enable Opus 5.5, GPT-6.1 Sol or GPT-6 Luna under model policies in Copilot settings, or your users lose those choices without a replacement.
  3. Pinned models in workflows and integrations. GitHub asks customers to update these to supported models before the date. A quick search of your repos and automation for the old model names is the fastest way to find them.
  4. Removal is automatic. You don’t need to do anything to take the old models out once they are deprecated.

Because spend now follows tokens, it is also worth checking your Copilot budgets after the switch, especially if some users move from Terra to Luna and others from Opus 4.8 to Opus 5.5.

Also on 15 October: two security features reach general availability

AI Scan for pull requests is now generally available. It looks for vulnerabilities in pull request changes in languages and frameworks CodeQL doesn’t cover yet. GitHub names PHP, Shell/Bash, Terraform (HCL), Dockerfiles, JSP and Blazor. Findings show up next to CodeQL results with an “AI” marker and can include an explanation and a suggested fix. Points to know:

  • It covers public repositories. Private and internal repositories owned by an organisation need GitHub Code Security or Advanced Security. A Copilot licence is not required.
  • Findings are advisory. They don’t fail the code scanning check and can’t block merges through rulesets.
  • It doesn’t run on pull requests from forks or on Dependabot pull requests.
  • Usage consumes Code Security AI Credits, which you can cap with SKU-level budgets.

Agentic autofix for code scanning alerts is also generally available for Code Security and Advanced Security customers on GitHub Enterprise Cloud, including data residency. You assign an alert to Copilot, or up to 25 alerts at once. It explores the codebase, proposes a fix, re-runs CodeQL to check that the alert closes, and opens a draft pull request explaining what it did.

Two details matter for budgets. Assigning an alert to Copilot replaces the free “Generate Fix” option. And usage draws on both AI Credits from your Code Security budget and GitHub Actions minutes. It only works on alerts on the default branch. An enterprise admin has to enable the policy first, an org or repo admin turns it on in Code Security settings, and a Copilot billing method must be set.

A short checklist

  • Before Monday 19 October: move anyone using Gemini 3.7 Flash, GPT-5.5, GPT-5.4, GPT-5.4 mini, GPT-5 mini or Grok 4.5.
  • Before 16 November: replace Claude Opus 4.8 and Opus 5 with Opus 5.5, and GPT-5.6 Terra with GPT-6.1 Sol (or Luna where it tests well).
  • If the global default is off in your org, enable the replacements yourself.
  • Search automation and agent configs for pinned model names.
  • Before switching from “Generate Fix” to agentic autofix, set a Code Security AI Credits budget.

References

  1. Upcoming deprecation of selected GitHub Copilot models in mid-November — GitHub Changelog, accessed 2026-10-17
  2. Upcoming deprecation of selected GitHub Copilot models in mid-October — GitHub Changelog, accessed 2026-10-17
  3. Models and pricing for GitHub Copilot — GitHub Docs, accessed 2026-10-17
  4. Model multipliers for annual plans on request-based billing (legacy) — GitHub Docs, accessed 2026-10-17
  5. AI Scan for pull requests generally available — GitHub Changelog, accessed 2026-10-17
  6. Agentic autofix for code scanning alerts general availability — GitHub Changelog, accessed 2026-10-17

This week in AI, every Sunday

One email with the releases that mattered and what they mean in practice. No spam.