AI Market Intelligence Daily

1. Largest movers — deadline compression dominates Kalshi. The most dramatic shifts were all on Kalshi's model-release markets, where short-dated contracts collapsed as deadlines approached: Grok 4.7 before Aug 14 crashed 83% (6% to 1%) and Gemini 3.5 Pro before Aug 16 fell 75% (8% to 2%). This is classic time-decay behavior — as the resolution date nears without an announcement, probability mass flows to later-dated contracts. Grok 4.7 before Aug 21 rose 57% (7% to 11%) as a beneficiary. Traders should watch whether this cascade continues: if Aug 21 also disappoints, the next expiry (Aug 28 at 20%) faces similar downward pressure.

2. OpenAI personnel markets reprice amid executive exodus. The Ari Weinstein departure market on Polymarket plummeted 53% (35% to 17%), but this likely reflects the Brad Lightcap departure news — OpenAI's longtime COO announced he's leaving to "start something new." The market appears to interpret Lightcap's exit as a structured, planned transition rather than chaos, reducing perceived instability and thus the probability that mid-level staff like Weinstein would also leave. However, Conrad Kramer's departure probability sits at 51% — essentially a coin flip — suggesting the market remains deeply uncertain about OpenAI's leadership durability. Combined with the 84% probability that OpenAI will NOT IPO by year-end, traders are pricing in organizational turbulence that delays any public listing.

3. Google Gemini news-price divergence warrants attention. Google announced the Pixel 11 lineup and new Gemini features at Made by Google '26, and Gemini hit 1 billion users — both bullish signals. Yet Gemini 3.5 Pro short-term markets are declining (Aug 16: down 75%, Aug 21: down 10%). The Aug 31 contract at 43% ticked up slightly (+5%), suggesting traders believe a Gemini 3.5 Pro release is still likely by month-end but won't come in time for the earlier deadlines. This divergence between positive product momentum and declining near-term release odds points to a gap between marketing announcements and actual product shipping timelines — a common pattern for Google, which tends to demo features at events long before general availability.

4. Anthropic roadmap confusion: Opus 6 rises while Mythos sinks. Opus 6 before Nov 1 surged 56% (9% to 14%), the second-biggest move of the day. Meanwhile, Mythos markets across all timeframes declined (Sep 1: down 50%, Dec 1: down 31%, Sep 1 Mythos-Class: down 19%). This split suggests the market is reallocating probability from the speculative "Mythos" branding toward the more concrete "Opus 6" label. Anthropic may be consolidating its messaging around the Opus line, making Mythos feel increasingly like vaporware. The 70% probability for "next Mythos-Class model before Oct 1" still looks overconfident given the broader Mythos selloff.

5. AI startup valuations run hot with little market friction. Cognition reportedly raising at $40B and Lovable at $13.3B with another $400M raise signal continued AI-coding startup euphoria. Yet none of these stories moved the Polymarket or Kalshi markets meaningfully — the prediction markets remain focused on model releases and corporate milestones rather than venture capital narratives. This disconnect suggests prediction markets are insulated from broader AI hype cycles, responding only to concrete events rather than funding announcements.