TL;DR
A circulating claim that Anthropic is 70% likely to release Claude Opus 4.8 and Sonnet 4.8 by May 31 appears to misread a Polymarket date ladder. The source material says May 31 was priced near 26%, while June 15 carried the roughly 70% figure, with no official Anthropic announcement.
A prediction-market claim circulating among developers appears to overstate the odds of an imminent Claude 4.8 release: source material from Thorsten Meyer AI says Polymarket showed roughly 26% odds for release by May 31, while the widely cited 70% figure applied to a later June 15 date.
The confirmed facts are narrower than the rumor. Anthropic has not announced Claude Opus 4.8 or Claude Sonnet 4.8, and the source material says there is no public API model ID, model card, benchmark release or official launch date for either model.
The current confirmed Claude 4 lineup cited in the source is Claude Opus 4.7, launched April 16, 2026, at $5 per million input tokens and $25 per million output tokens with a 1 million token context window; and Claude Sonnet 4.6, launched February 17, 2026, at $3 per million input tokens and $15 per million output tokens, also with a 1 million token context window.
The rumor rests partly on a March 31 Claude Code source-map reference to a sonnet-4-8 string inside what the source describes as a forbidden-model-names filter list. That string is evidence that the name appeared in code, but it does not confirm that Anthropic will ship a model under that name or that a release is scheduled.
Why It Matters
The distinction matters because developers and AI buyers may delay model, tooling or agent-workflow decisions if they believe a new Claude release is days away. The market signal described by the source points to a plausible near-term release, not a confirmed end-of-May launch.
The thin trading volume also limits what the odds can show. The source cites about $900 in total Polymarket volume as of May 27, 2026, making the market more useful as a sentiment gauge than as evidence of Anthropic’s internal release plan.
For teams building with Claude today, the practical takeaway is that Opus 4.7 and Sonnet 4.6 remain the available models. Any claims about Sonnet 4.8 vision performance, software-engineering benchmark gains or enterprise impact are forecasts unless Anthropic publishes details.
MCP Server Security Fundamentals: How to secure MCP servers, tool access, credentials, permissions, and exposed endpoints
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background
The source frames the Polymarket event as a date ladder, with separate odds for separate deadlines. Read that way, the market suggests increasing confidence over time rather than one single probability attached to May 31.
One pattern detail complicates the rumor. The source says Anthropic has paired Opus and Sonnet minor versions consistently, but there has been no Sonnet 4.7 release. A direct move from Sonnet 4.6 to Sonnet 4.8 would break that pattern, although it could also indicate an attempt to re-align Opus and Sonnet naming.
The rumor has reportedly been circulating for about eight weeks. During that period, the source says there has been no official Anthropic signal confirming Opus 4.8 or Sonnet 4.8.
“The 70% number is real.”
— Thorsten Meyer AI
“May 31 is actually ~26%.”
— Thorsten Meyer AI
“No Anthropic announcement, no API id, no model card, no benchmarks for 4.8.”
— Thorsten Meyer AI
The Mathematics of Large Language Models: From Tokens to Transformers, Training to Decoding
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Remains Unclear
It is not yet clear whether Claude Sonnet 4.8 or Claude Opus 4.8 will ship, when they might ship, or whether either model would use those names. It is also unknown whether the leaked string reflects a planned product, an internal placeholder, a blocked alias or an unused reference.
The prediction market does not confirm Anthropic’s roadmap. Its low volume means a small number of traders can move the displayed odds, and the source does not cite any direct statement from Anthropic about a Claude 4.8 launch.
![Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results](https://m.media-amazon.com/images/I/415+fSJacsL._SL500_.jpg "The 4.8 Staircase: What the Market Actually Believes About Claude’s Next Release 5")
Claude AI for Beginners Bible: [5 in 1] The Ultimate Guide to Automate Your Work, Save Hours Every Week, and Use AI for Real-World Results
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What’s Next
The next firm signal would be an Anthropic announcement, API listing, model card, documentation update or benchmark publication. Until then, readers should treat the May 31 release claim as unconfirmed and track date-specific market odds separately rather than as one headline number.
Evals for AI Engineers: Systematically Measuring and Improving AI Applications
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Is Claude 4.8 confirmed?
No. The source material says Anthropic has not announced Claude 4.8 and that there is no public API ID, model card or benchmark release for it.
Was the market really pricing 70% odds?
Yes, according to the source, but not for May 31. The roughly 70% figure was tied to June 15, while May 31 was cited near 26%.
What evidence supports the rumor?
The main cited evidence is a March 31 Claude Code source-map string, sonnet-4-8, plus a rising Polymarket date ladder. Neither confirms a release.
What Claude models are available now?
The source lists Claude Opus 4.7 and Claude Sonnet 4.6 as the confirmed current models, both with 1 million token context windows.
Why does the missing Sonnet 4.7 matter?
The source says Anthropic has previously paired Opus and Sonnet minor versions. A move from Sonnet 4.6 to Sonnet 4.8 would break that pattern, which makes the naming claim less settled.
Source: Thorsten Meyer AI
