Skip to content
NewsProduct launch

GitHub Copilot adds all three GPT-5.6 tiers to its model picker

· by Pondero Newsdesk

The short version

GitHub added GPT-5.6 Sol, Terra, and Luna to Copilot on July 9. Sol is gated to Pro+, Max, Business, and Enterprise; Terra and Luna are available on all paid plans including the base Pro tier.

GitHub Copilot adds all three GPT-5.6 tiers to its model picker

GPT-5.6 Sol arrived in GitHub Copilot on July 9 with a plan restriction that forces a choice for any team on the base Pro plan: keep the model you have or upgrade to reach the new ceiling.

What happened

GitHub's changelog confirmed that all three GPT-5.6 variants are now selectable inside GitHub Copilot. The three tiers carry distinct roles per the changelog:

  • GPT-5.6 Sol targets complex reasoning over large codebases and demanding, long-running agentic tasks. It is limited to Copilot Pro+, Max, Business, and Enterprise plans.
  • GPT-5.6 Terra is positioned as the balanced default for everyday interactive and agentic coding. It is available on all paid plans, including Pro.
  • GPT-5.6 Luna is the lightweight, lower-cost variant for smaller or faster tasks and sits at the lowest price point in the family. It is also available on all paid plans.

All three are billed at provider list pricing under GitHub Copilot's Usage Based Billing, per the changelog. The models are selectable in VS Code, Visual Studio, JetBrains, Xcode, Eclipse, the Copilot CLI, the GitHub Copilot cloud agent, the Copilot app, GitHub Mobile, and github.com. GitHub's docs list Claude Fable 5 and Kimi-K2.7-Code as other models already in the same picker.

Rollout is gradual; the changelog notes that users who do not see the models yet should check back. For Business and Enterprise plans, administrators must turn on the GPT-5.6 policy in Copilot settings; it is off by default.

Why it matters

The Sol restriction is the practical decision point. Any team on Copilot Pro today can access Terra and Luna immediately. Sol, the highest-capability tier in the family, requires at least Pro+ or a Business or Enterprise seat. If your use case involves extended agentic runs or reasoning over large codebases, that is the tier GitHub and OpenAI are pointing to, and it sits behind a plan gate.

The tier split also gives cost-conscious teams a reason to stay on a lower plan. Luna is explicitly the lower-cost option. Teams that run short, repetitive code tasks or use Copilot mostly for completions can use it without stepping up to the more expensive Sol.

For Copilot admins, the default-off policy for Business and Enterprise has an immediate practical implication. The models do not appear for your team until you enable them in settings. If you want developers testing GPT-5.6 Sol on agentic tasks right now, the toggle needs to happen before any benchmark comparisons against Fable 5 or Kimi-K2.7-Code are meaningful.

What to watch next

GitHub has not published a side-by-side coding benchmark comparing the three GPT-5.6 tiers against the existing field inside Copilot. That comparison matters most for teams deciding whether Sol justifies the Pro+ or Business seat cost over Terra alone. Watch GitHub's documentation and community forums for early performance reports. The changelog also references a GPT-5.6 Sol Pro variant in OpenAI's plan materials; whether that reaches Copilot, and at what plan tier, is still open.

Sources