Anthropic appears close to releasing Claude Opus 5, with preparations spotted among the company's partners pointing beyond internal testing.

The clearest signal emerged earlier this month when a research entry briefly surfaced in a coding tool's model picker before being removed within hours. The listing showed an early access preview featuring per-turn controls, safety fallbacks, a one million token context window, and an extra-high-effort setting.

A matching model string subsequently appeared in Google's Vertex quotas catalog, following the same pattern that preceded Sonnet 5's release.

No pricing, system card, or public model identifier exists yet, leaving the specification list provisional.

The model would likely appear across Claude's ecosystem — the consumer apps for paid tiers, Claude Code, the Claude Platform, and the three cloud partners. It would most likely replace Opus 4.8 rather than running alongside it.

The safety fallback detail proves telling: routing flagged prompts down to version 4.8 mirrors the arrangement used above the Opus tier, suggesting this release targets frontier-level performance despite its numerical designation.

Why positioning matters for Anthropic

Opus has faced pressure from both directions. Sonnet 5 arrived in late June performing close to 4.8 at significantly lower cost, while Fable and Mythos models sit above it in capability.

Subscription-inclusive Fable 5 access ended on 19 July, replaced by usage credits. A stronger Opus would give Max, Team, and Enterprise customers an upgrade path without paying Mythos rates.

Release cadence supports the timing speculation:

  • 70 days from version 4.6 to 4.7
  • 42 days from 4.7 to 4.8
  • 56 days elapsed since 4.8

Expectations center on improved coding performance, though Sonnet 5's mixed reception demonstrates that user expectations aren't always met.

Anthropic has not commented on the potential release.