Data: Individual stock ETPs listed on South Korea's stock market have caused retail investors to suffer losses exceeding 6 trillion won in two and a half months.

Insight Beating AI Flash News: OpenAI’s Product and Platform Lead Thibault Sottiaux announced Day 16 of the company’s 28-day product sprint, with a key focus on resolving context caching issues for long-running Agents. For logged-in Codex and ChatGPT Work users, OpenAI has announced permanent retention of prompt caches. This ensures cache hits even when resuming work after hours or reopening old sessions, eliminating the loss of cache reuse benefits due to long pauses. The API has also been updated: OpenAI extended the optional cache duration for the Responses API from the default 30 minutes to 12 hours. Developers can cache system instructions, tool definitions, and common context long-term, so Agents can reuse prior computation results even when resuming tasks hours later. The two durations can be mixed: fixed instructions are retained for 12 hours, while new conversation content is cached for only 30 minutes. However, extending API caching incurs additional costs. The initial write price for 12-hour caching is 2.5 times that of standard input, double the cost of the default 30-minute cache. For subsequent cache hits, input prices for most new models are reduced by 90%, and by 95% for GPT-6.1 Sol. Thus, long caching is more suitable for Agents that repeatedly call the same context, rather than short, one-off tasks.

Beating AI News Flash: OpenAI’s Product and Platform Head Tibo Sottiaux admitted that some users have recently encountered abnormal Codex usage quota consumption. An investigation found that roughly 40,000 accounts have actual usage limits inconsistent with their selected subscription plans. OpenAI has fixed the issue, is contacting affected users individually, and has committed to reissuing usage credits.

NEAR’s governance platform House of Stake has opened voting on the HSP-029 proposal, which seeks to gradually lower the NEAR token’s annual issuance rate from the current 2.5% to 1.6% over the next 24 months, to reduce token dilution and optimize the long-term token economic model. The voting deadline is November 2, 2026.

Super Micro Computer raised its preliminary quarterly revenue and gross margin guidance, driving Dell (DELL)’s post-market share price to rise around 6% at one point. According to TradingBeats monitoring, the largest DELL long position holder, who had been averaging down during the prior decline, reduced their position by roughly $960,000 in the early hours of today amid the current rebound, nearly half of their original holding. The total loss on the reduced position was around $6,300, exiting nearly at break-even. Currently, the whale still holds 2,200 DELL long contracts, with a position value of approximately $1.08 million, an average entry price of around $482, uses 4x isolated margin leverage, and has an unrealized profit of about $17,000 on the remaining position.

Solana to Launch Alpenglow Upgrade on October 21, One of Network’s Largest Underlying Overhauls Since 2020 Solana will roll out its Alpenglow upgrade on October 21, marking one of the most significant foundational overhauls the network has undergone since 2020. The upgrade will replace the existing validator voting mechanism, aiming to cut transaction finalization time from roughly 6 seconds to 150 milliseconds—an approximate 40x speed boost. At the core of the upgrade is shifting validator voting from on-chain to off-chain, consolidating votes into certificates that enable transaction finalization within 1 to 2 consensus rounds. If mainnet performance meets expectations, Solana’s transaction confirmation speed will approach that of credit card payments or centralized exchanges, enhancing user experiences for meme coin trading, stablecoin payments, and on-chain platforms. The upgrade will also reduce validator costs. Currently, validators pay around 2 SOL in daily voting fees. Post-upgrade, voting fees will be eliminated, replaced by a fixed validator access fee per epoch: 0.8 SOL per ~24-hour epoch, an estimated cost reduction of roughly 60%. However, the testing process faced hurdles. According to Galaxy Digital, since the testnet launched in early May, it has been wiped and restarted from scratch six times. One migration attempt caused a full testnet failure requiring a rollback; subsequent migrations ultimately succeeded, paving the way for the mainnet launch. All issues occurred on the testnet, with no impact on real funds. Per the announcement, the upgrade will begin at the start of Epoch 1064 on Wednesday, and complete the switch at Slot 459,653,000 after a 5,000-slot transition phase.

Dragonfly Management Partner and early Polymarket investor Haseeb Qureshi posted that claims Polymarket CEO Shayne Coplan was caught using company funds to cover private jet flights and other expenses are "total nonsense" and an unsubstantiated smear report. Qureshi added that Coplan founded Polymarket at age 21, took no salary at the time, ran the company from an apartment and lived frugally. Dragonfly was one of the first firms to extend an investment term sheet to Coplan, and even proposed a personal loan to him. Coplan disclosed all his expenses, with the annualized total roughly equivalent to a software engineer’s salary; the flight costs in question did not involve personal travel. Qureshi also emphasized that Coplan voluntarily initiated the accounting review and refund process, not after investors discovered the issue and demanded repayment. Coplan has since fully reimbursed the company, and now receives a regular salary like other employees. Qureshi characterized the matter as an accounting error, not an integrity lapse, and stated that Dragonfly retains full confidence in Coplan and his leadership. He noted that Polymarket has grown from a small startup to one of the fastest-growing and most impactful consumer brands in the U.S., and called on the public to stop spreading the negative reports.