RUMOR

If the tip holds up, it strikes right at the most criticized part of Opus 5: its reputation for rewriting whole files without being asked, stalling on unnecessary confirmations and losing the thread on long tasks, a pattern the community has nicknamed opusfived. An identifier reading claude opus 5 2 reportedly showed up in the Microsoft AI Foundry catalog, and users claim Claude Code is already silently routing part of Opus 5 traffic to this new model.

The tip that lit the rumor

The entire story runs through a single X account, Mr Salio, who posted a chain of claims: the new model reportedly beats Fable 5.1 in internal tests, is being tested inside Claude Code, Anthropic is said to be skipping straight from Opus 5 to 5.2 without a 5.1 in between, and the claude opus 5 2 slug has reportedly appeared in Microsoft's Foundry, the channel companies use to reach Anthropic models through Azure. None of these claims came with a screenshot of official documentation, a changelog entry or a model card.

Single source, multiple claims

As of this morning, Anthropic's public documentation still lists Claude Fable 5.1, Claude Opus 5, Claude Sonnet 5 and Claude Haiku 4.5 as the current models. There is no changelog entry, status page or cloud catalog line, whether on OpenRouter, Bedrock, Vertex or Foundry itself, publicly citing an Opus 5.2. A second report, from another account, points the same direction with a separate claim: a specific prompt about a user known as the reset guy reportedly answered differently in Claude Code versus Devin, which the author read as evidence of routing to a new model inside Claude Code.

Why the prompt test doesn't settle it

A different answer between two tools does not isolate the model as the variable. System prompt, effort level, tool configuration and even normal sampling variance produce different outputs from the same model across different interfaces. It is anecdotal evidence that something changed, not a controlled comparison proving a new model exists, let alone that it is called Opus 5.2.

Why it's still worth watching

Anthropic has been shipping on short cycles since Fable 5.1, and competitive pressure from GPT 6 Astra plus the swirling rumors around a Fable 5.2 give strategic context for the company to also reinforce the Opus line, currently the most criticized model in its catalog for slowness and overengineering on simple tasks. If Anthropic wants a fast answer to Opus 5 complaints, evolving that line before touching the top of the table with Fable makes product sense.

MaxAssistant's read

This is a single source rumor with no verifiable technical artifact and a routing claim built on a test that doesn't isolate variables. Treat it as a community backroom bet, not a launch in progress. The real turning point isn't another post on X, it's the claude opus 5 2 identifier showing up documented on Anthropic's own site, in a model card or on an official cloud provider page.

Sources

Mr Salio on X, Opus 5.2 leak thread: https://x.com/Mr_Salio/status/2099830955947049075 | CellCog, Claude Opus 5.2 Release Date, The Routing Rumor Dated: https://cellcog.ai/blog/claude-opus-5-2-release-date/ | Kie.ai, Claude Opus 5.2 101, The Claimed Jump From Opus 5: https://kie.ai/blog/what-is-claude-opus-5-2