GitHub Copilot suspends new sign-ups and restricts advanced models

GitHub Copilot, the developer-facing coding assistant, paused new sign-ups for individual plans and imposed tighter access controls on its top-tier AI models after unprecedented usage growth driven by autonomous agents. The company removed the Opus model line from the standard Pro plan, moved Claude Opus 4.7 behind the more expensive Pro+ tier, and announced that Opus 4.5 and Opus 4.6 will be removed even from Pro+.

The enforcement includes two operational limits, a session-level cap to smooth peak-hour load and a hard weekly token budget that, if exceeded, removes access to premium models and automatically routes the account to a fallback set of lower-capability models until the window resets. IDE integrations such as VS Code and the Copilot CLI will display remaining quota to reduce surprise interruptions. The changes target heavy multi-request workflows, including orchestrated agents and subagents that generate continuous, high-volume token consumption. These requests have sometimes produced usage costs far exceeding a single subscription fee, creating unsustainable tail usage.

This is a pragmatic operational response to rapid behavioral change among developers. As teams adopt autonomous-agent patterns for long-running tasks and pipeline automation, provider economics shift from per-seat flat subscriptions to high-variance token consumption. GitHub is aligning product tiers with marginal cost by gating the highest-capacity models behind a premium plan and adding deterministic caps to protect latency and availability. The move echoes similar access controls elsewhere in the ecosystem as model-serving costs and cold-starts make unbounded usage impractical.

Developers who rely on agentized workflows will need to evaluate plan upgrades or implement local/offline execution to avoid throttling. Watch for follow-on changes: metered pricing options, explicit token-based paid tiers, or partner integrations that offer reserved capacity for heavy users. Also monitor whether other platform providers adopt similar session and weekly token caps to contain outlier usage.