Tom Lee: Wash’s remarks are more dovish, which bodes well for the recovery of risk assets and AI trading.

After the release of its Q3 report, Citizens downgraded Alphabet from "Outperform" to "Market Perform" and did not set a price target. The firm noted that while Alphabet’s query growth hit a five-year high, its monetization faces challenges, and its search business is dealing with tough year-over-year comparisons. Citizens believes effective monetization transition from AI Overviews to agent tools requires strong execution. With intensifying competition in agent tools, the firm points out that Google faces "significant execution risks during the search monetization transition process".

JPMorgan analyst Brian Essex has initiated coverage on Palantir, assigning an Overweight rating and a $262 price target, implying more than 30% upside from the current share level. The bank has also added the stock to its "Positive Catalyst Watch List" ahead of Palantir's earnings release on November 2. The firm noted that as enterprises explore AI applications, Palantir has delivered return on investment by driving fundamental changes to clients' core business operations via its platforms. The analyst told investors in the research report that the company's growth is accelerating amid rising demand for AI sovereignty. JPMorgan added that Palantir enables enterprises and government organizations to secure and own their own production resources and services.

According to an official announcement, Binance's derivatives platform will launch USDT-margined perpetual contracts at the following times on October 29, 2026 (UTC+8): iShares MSCI Brazil Small-Cap ETF (EWZS) at 18:05, NuScale Power (SMR) at 18:10, Occidental Petroleum (OXY) at 18:15, Amgen (AMGN) at 18:20, Pfizer (PFE) at 18:25, and MercadoLibre (MELI) at 18:30. All contracts support a maximum leverage of 20x.

Insight Beating AI Flash News: Suhail Doshi, co-founder of product analytics platform Mixpanel and founder of AI image platform Playground, has launched a new venture called Particle, debuting its first model, Particle-1. Trained via reinforcement learning on DeepSeek V4.1 Flash, the model is designed to reduce token consumption during coding agent tasks. Four programming and data engineering benchmarks released by Particle show that Particle-1 cuts output tokens by 28.4% to 52.2% compared to the original DeepSeek model, with an overall average reduction of around 45%, while task pass rates stay within the original model’s statistical confidence interval. Estimated at the same token unit price and fixed cache hit rate, per-task costs drop by 34.4% to 51.6%. The token savings stem mainly from more streamlined inference, fewer retries, and early termination after task completion. In a Git deployment test, both models successfully finished the task: Particle-1 output only 15,892 tokens, while the original model used 47,006 tokens. Particle-1 has launched an open API and integrated with Vercel AI Gateway. Current performance and cost data are primarily from the team’s internal testing.

According to U.S. financial media Semafor, the Trump administration is in ongoing negotiations over a temporary peace agreement between the U.S. and Iran.

Bitget market data shows WTI crude oil has dropped below $100 per barrel, logging an intraday decline of 1.23%. Spot gold has rallied in the short term.