Claude Opus 5.5 became selectable in GitHub Copilot on September 22, 2026. GitHub lists Copilot Pro+, Max, Business and Enterprise as eligible plans, across ten Copilot surfaces. The rollout is gradual, and organization administrators can control access. Under current billing, the model uses GitHub AI Credits at Copilot’s published token rates; GitHub’s legacy annual Pro and Pro+ plans are an exception and do not receive new models. GitHub’s launch notice and its current billing pages are the controlling sources for access and cost.
This guide covers the Copilot integration: where the model appears, which plans can select it, how usage is charged, and what to check if it is missing. Anthropic’s wider Claude Opus 5.5 launch is a separate model-selection and API story; Copilot’s AI-credit prices below are not Anthropic API prices.
Which Copilot plans include Opus 5.5?
GitHub names four eligible plans: Pro+, Max, Business and Enterprise. Copilot Pro, Free and Student are not listed as selectable plans for this launch; Free and Student are documented as Auto-model-selection-only. Check the account’s billing mode as well as its plan: existing annual Pro or Pro+ subscribers who stayed on legacy premium-request billing do not get access to new models and features.
| Plan | Published plan price | Monthly AI-credit allowance | Opus 5.5 eligibility in GitHub’s launch notice |
|---|---|---|---|
| Copilot Pro | $10/month | 1,500 total | Not listed |
| Copilot Pro+ | $39/month | 7,000 total | Eligible, unless on the legacy annual request-billing path |
| Copilot Max | $100/month | 20,000 total | Eligible |
| Copilot Business | $19 per granted seat/month | 1,900 per licensed user, contributed to a shared organization pool | Eligible; an administrator’s model policy can restrict it |
| Copilot Enterprise | $39 per granted seat/month | 3,900 per licensed user, contributed to a shared organization pool | Eligible; an administrator’s model policy can restrict it |
| Copilot Free or Student | Free | Allowance varies by plan | Not named as eligible; plan docs say Auto model selection only |
Individual-plan totals above include the base and flex credit amounts shown in GitHub’s plan table. Business and Enterprise credits are pooled at the billing entity, so the per-user figures are contributions rather than isolated personal wallets. GitHub says extra use is billed in AI Credits; administrators should review the organization budget and paid-usage settings before enabling a high-volume model.
How much does it cost in Copilot?
GitHub converts metered model usage into AI Credits at the published rates. One credit is $0.01. The current Copilot table lists these Opus 5.5 rates per one million tokens:
| Token category | Copilot list rate per 1M tokens |
|---|---|
| Input | $4.00 |
| Cached input | $0.20 |
| Cache write | $5.00 |
| Output | $20.00 |
For example, suppose one interaction is billed as 100,000 input tokens, 20,000 cached-input tokens, 10,000 cache-write tokens and 5,000 output tokens. At the rates above, that is $0.400 + $0.004 + $0.050 + $0.100 = $0.554, or 55.4 AI Credits, before any interface rounding. This is a worked calculation from the published rates, not a measured Copilot session. Real agent work spans multiple interactions, and its total depends on the tokens Copilot reports in each category.
Anthropic models have a separate cache-write rate in GitHub’s table. Do not price a cache write as either ordinary input or a cached read. The same model’s direct Anthropic API has its own rate card and billing relationship; this article’s arithmetic applies only to GitHub Copilot usage-based billing.
New usage-based billing replaced the old premium-request system on June 1, 2026. It is token-based: the model and number of tokens determine the charge. The legacy request multipliers remain relevant only to certain existing annual Pro and Pro+ subscriptions. GitHub explicitly says legacy annual users do not receive new models, so a Pro+ label alone does not establish Opus 5.5 access. See GitHub’s billing transition guide and legacy-plan rules.
Where can you use it?
GitHub’s September 22 notice lists Claude Opus 5.5 in the model picker for these ten surfaces:
- Visual Studio Code and Visual Studio
- Copilot CLI
- GitHub Copilot coding agent and the Copilot app
- github.com
- GitHub Mobile on iOS and Android
- JetBrains IDEs, Xcode and Eclipse
The notice says rollout is gradual. GitHub’s supported-model directory currently marks Opus 5.5 generally available, but that status does not mean every account or picker received the change at the same moment. The docs list Visual Studio 17.14.6 as a model-specific minimum and show “TBD” for several other clients; GitHub cautions that an unspecified minimum is not proof that every older version works. Update the current Copilot client or extension before troubleshooting.
Why might Opus 5.5 be missing?
- Confirm the account’s plan and billing path. Pro+ is named as eligible, but legacy annual Pro and Pro+ subscriptions on request-based billing do not receive new models.
- Allow for the gradual rollout. GitHub says to check again if the model is not yet visible.
- Ask an organization administrator to check model policy. Business and Enterprise administrators can control the setting. New models are enabled by default only when the organization has not disabled global default model enablement or explicitly blocked the model.
- Update the client. Use the current supported-model documentation for IDE and extension guidance. “TBD” means no minimum was listed in the reviewed release guidance, not that all old versions are supported.
GitHub’s supported-model page still lists Claude Opus 4.7 as retiring on October 2 and names Claude Opus 5 as its suggested alternative. That table does not name Opus 5.5 as the replacement, so do not infer that this launch changes the migration recommendation. For the existing Copilot model-deprecation and admin checklist, see our September Copilot guide.
What this announcement does and does not establish
GitHub describes Opus 5.5 as suitable for agentic coding, long-running agent tasks and knowledge work, and reports favorable early internal testing. Those are vendor statements, not an independent benchmark or a guarantee about a team’s tasks. This access guide does not rank Opus 5.5 against other models or claim that it will reduce a user’s token spend. Use a representative task and your Copilot usage data to decide whether its results justify the observed credits.
For model-wide capabilities, direct API details and Anthropic’s own pricing, read the separate Anthropic Opus 5.5 announcement and current platform documentation. Copilot plan eligibility, rollout, organization controls and credit rates can change independently of that direct API offer.
Access and billing details reflect GitHub documentation checked September 23, 2026. Rollout may still be in progress and organization policy can change what an eligible user sees. No Copilot account was tested and no independent performance benchmark was run for this guide.
Sources
- GitHub: Claude Opus 5.5 is now available in GitHub Copilot
- GitHub Docs: Supported AI models in GitHub Copilot
- GitHub Docs: Plans for GitHub Copilot
- GitHub Docs: Models and pricing for GitHub Copilot
- GitHub Docs: What changed with Copilot billing
- GitHub Docs: Model multipliers for annual plans on legacy billing
- Anthropic: Introducing Claude Opus 5.5
