- Anthropic’s Claude Code model configuration support page lists Opus 4.7 (
claude-opus-4-7) as the newest supported model, above Sonnet 4.6 and Opus 4.6. - Hacker News users reported on April 28, 2026 that Claude Pro subscribers now need to enable an extra-usage setting before Opus-tier models are accessible in Claude Code — a claim not confirmed by a standalone Anthropic policy announcement at time of writing.
- The support document confirms three switching mechanisms: the interactive
/modelcommand, the--modelsession flag, and the persistentANTHROPIC_MODELenvironment variable. - The extra-usage gating claim originates from user reports, not from the primary source documentation itself; the sourcing distinction matters for interpreting the story.
What Happened
Anthropic’s Claude Code model configuration support page now lists Opus 4.7 — model identifier claude-opus-4-7 — as the top entry in its roster of selectable models. Separately, Hacker News users flagged on April 28, 2026 that Claude Pro subscribers are now required to enable an extra-usage toggle before Opus-tier models become accessible within Claude Code sessions. Anthropic had not published a standalone policy announcement documenting that access change at the time of writing.
Why It Matters
If accurate, the extra-usage requirement would represent a structural change to what Claude Pro’s flat subscription rate delivers: previously, Pro users could route Claude Code sessions to any supported Opus model without additional account-level steps. The practice of gating frontier model access behind extra toggles or higher usage tiers is established at OpenAI (o-series reasoning models behind “Advanced” settings) and Google (Gemini Ultra behind Workspace add-ons), making the reported shift consistent with broader industry tiering patterns.
Technical Details
Anthropic’s support documentation confirms six models currently selectable within Claude Code: Opus 4.7 (claude-opus-4-7), Sonnet 4.6 (claude-sonnet-4-6), Opus 4.6 (claude-opus-4-6), Opus 4.5 (claude-opus-4-5-20251101), Haiku 4.5 (claude-haiku-4-5-20251001), and Sonnet 4.5 (claude-sonnet-4-5-20250929). The document describes three configuration pathways: the /model interactive slash command for immediate in-session switching, the --model flag at launch for single-session overrides, and the ANTHROPIC_MODEL environment variable written to ~/.zshrc or ~/.bashrc for persistent defaults. The documentation notes that “the simplest way to change models is to use the /model command directly within Claude Code” and that model changes “take effect immediately” without restarting the terminal. Users can verify the active model at any time via the /status command.
Who’s Affected
Claude Pro subscribers who route Claude Code sessions to Opus-class models for complex reasoning or long-context tasks would be the most directly affected cohort if the extra-usage gate is confirmed. Enterprise and API customers billed per token are governed by usage-based pricing rather than subscription caps, so the access friction — if real — would not apply to that segment. Developers managing shared team environments via the ANTHROPIC_MODEL environment variable would also need to verify that each team member’s account has the required toggle activated at the account level, since a shell-level default cannot override a subscription-tier restriction.
What’s Next
Anthropic has not issued a formal changelog or product blog post documenting the reported access change as of April 28, 2026; the company’s support page focuses exclusively on configuration mechanics rather than subscription policy. Users who encounter Opus access issues in Claude Code should check their account’s usage settings for an extra-usage or model-access toggle, and monitor the Claude Code model configuration support page for policy language additions. Whether the Opus 4.7 listing in the documentation reflects a fully shipped release or a staged rollout also remains unspecified in the available source.
