Anthropic faces 91% odds of Claude Opus launch by end of July

Prediction markets give Anthropic a 91% chance of shipping its next Claude Opus model by the end of July, with odds of a launch this week surging 36 points in recent days.

"Opus 4.8 was a bridge rather than a ceiling," Dario Amodei, Anthropic's chief executive, said in April, conceding the model trailed the company's unreleased Mythos-class systems on internal benchmarks.

The Polymarket contract, which has attracted more than $227,000 in trading volume, assigns a 68% probability to a launch by Thursday, 24 July, and a 95% chance by the end of August. Odds jumped sharply on 1 July, when the December contract moved from around 50 cents to 93 cents in a single session, suggesting traders had seen a corroborating signal. Rumored capabilities include major gains in agentic coding, context windows above 1 million tokens and improved efficiency, though none has been confirmed by the company.

A new Opus would arrive as Chinese rival Moonshot AI's Kimi K3 — the largest open-weight model ever released — tops leaderboards for frontend coding and ranks third on Artificial Analysis's Intelligence Index, behind only Anthropic's own Claude Fable 5 and OpenAI's GPT-5.6 Sol. The release tests whether Anthropic can maintain its lead at the frontier as open-source alternatives narrow the gap.

Anthropic has shipped a new Opus flagship roughly every six to eight weeks this year, a cadence that has fed speculation that a successor — potentially a full Claude Opus 5, with the internal codename Honeycomb — is close. A separate Polymarket contract asking whether the model lands specifically on Thursday, 23 July trades at 29%, while traders assign a 95% chance that a new Claude Sonnet model will arrive by the end of December and 51% for a new Haiku.

The Competitive Pressure Intensifies

The timing is significant. Moonshot AI's Kimi K3, released on Thursday, topped Arena's frontend coding leaderboard within a day, ahead of every leading US model, and placed third on Artificial Analysis's Intelligence Index — behind only Anthropic's Claude Fable 5 and OpenAI's GPT-5.6 Sol. Moonshot, backed by Alibaba and Tencent and valued at roughly $31.5 billion, says K3 still trails the most powerful proprietary models overall but beats second-tier systems including Claude Opus 4.8 and GPT-5.5 on benchmarks for coding and agentic tasks.

The release drew sharp reactions from US tech leaders. David Sacks, cochair of the President's Council of Advisors on Science and Technology, called the development "concerning," arguing that US regulatory overreach risks ceding the AI lead. Vinod Khosla, founder of Khosla Ventures, agreed, pointing to immigration policies that he said are scaring away foreign talent. Gavin Baker, chief investment officer at Atreides Management, called Kimi K3 an "inflection point," saying anything that lowers margins at the model layer benefits every other layer of the AI stack.

For Anthropic, the stakes are clear. The company's rapid release cadence has kept it at the frontier, but each new model faces diminishing differentiation as open-weight alternatives improve. A Claude Opus 5 that delivers on rumored capabilities — particularly in agentic coding and extended context windows — could widen the gap. But the window for proprietary advantage is narrowing: Kimi K3's performance on coding benchmarks, achieved at a fraction of the valuation of Anthropic or OpenAI, suggests the cost of frontier capability is falling faster than the market has priced in.

This article is for informational purposes only and does not constitute investment advice.