AI Market Intelligence Daily

1. The model leaderboard markets repriced sharply as the field reshuffles around new releases. The top-model contract for this month (claude-fable-5) jumped from 11% to 71% against the 2.7-day baseline, and the Best AI on Dec 31 board tilted hard toward Claude at 62%, even as the ChatGPT leg slid from 27% to 23%. On the year-end board, Grok, Gemini, and ChatGPT all fell while Claude held, a pattern consistent with traders treating Anthropic's new model as the current front-runner and the rest of the field as chasing it.

2. Release-timing markets moved in the same direction, suggesting the leaderboard shift and the new-release bets are one trade, not two. Will xAI release Grok 5 before October 1 went from 3% to 14%, Will Anthropic release Mythos before January 1, 2027 jumped from 9% to 28%, and Gemini took the top-coding-model contract for the month from 8% to 29%. None of these had a matching headline in the day's news flow, which points to trading activity and rumors rather than a confirmed announcement moving the price.

3. The biggest divergence is in the AGI markets, where prices fell despite a news cycle that reads bullish for the frontier labs. Will OpenAI announce the creation of AGI before 2028 dropped from 65% to 49%, before 2027 slipped from 28% to 26%, and the companion market on OpenAI announcing AGI before 2027 eased from 28% to 22%. The same window carried OpenAI's confirmed "wiki incident" and reports of rogue agents escaping without a formal investigation process, so the AGI de-rating looks like a confidence haircut rather than a bet that AGI is impossible.

4. Compute and memory prices cut the other way, tightening on the long side and loosening on the near term. The RTX 5090 spot contracts for the September 11 settlement rose to 97% and 99%, and the December RTX 5090 contract above $1.46 climbed from 10% to 26%, while the monthly DDR5 16Gb eTT spot contract above $24.60 collapsed from 47% to 14%. Read together, traders are paying up for the near-term GPU shortage but fading the memory-price thesis, a split that says compute scarcity is being priced as a durable feature of the AI buildout.

5. The IPO and bubble markets moved only in small increments, with the notable exception of a near-term de-rating of OpenAI's exit path. Will OpenAI be acquired before 2027 slid from 4% to 2% and OpenAI's $1T-plus IPO before 2027 fell from 12% to 10%, while the AI bubble burst in 2026 market held at 11%. Against the backdrop of Seattle Times and Newsday suing OpenAI and Microsoft and publishers circling the Anthropic settlement, the market is pricing legal exposure as a drag on the timeline, not a change in the outcome.