Home > Quick > Body

OpenAI员工公开指责xAI最新AI模型Grok3的基准测试结果具有误导性

clock
2025-02-23 02:23:53
近日,OpenAI的一名员工公开指责埃隆・马斯克旗下的xAI公司,称其发布的最新AI模型Grok3的基准测试结果具有误导性。对此,xAI的联合创始人伊戈尔・巴布什金(Igor Babushkin)则坚称公司并无不当。xAI的图表显示,Grok3的两个版本——Grok3 Reasoning Beta和Grok3 mini Reasoning——在AIME 2025上的表现超过了OpenAI当前最强的可用模型o3-mini-high。然而,OpenAI的员工很快在X平台上指出,xAI的图表并未包含o3-mini-high在“cons@64”条件下的AIME 2025得分。巴布什金在X平台上辩称,OpenAI过去也曾发布过类似的误导性基准测试图表。尽管这些图表是用于比较其自身模型的表现。
Web3 Desktop Trading Tool
Stay ahead of the game in the cryptocurrency space.

7x24 Newsflash

00:28 2025-06-18
JPMorgan Chase executives meet with SEC to discuss the process of capital markets chain
According to Cointelegraph, JPMorgan Chase executives recently met with the Securities Exchange Commission (SEC) cryptocurrency task force to discuss the potential far-reaching impact of digital asset regulation and the blockchain of capital markets. According to an SEC note released on Tuesday, JPMorgan executives spoke with the SEC about "the possible impact of migrating existing capital markets activities to public blockchains".
00:16 2025-06-18
Market news: U.S. plans to ease capital restrictions on bank bond transactions
Market news: The United States plans to ease capital restrictions on bank bond transactions. (Jin Ten)
00:16 2025-06-18
Three giant whales spent nearly $6 million hoarding HYPE
According to Onchain Lens monitoring, three giant whale accounts have recently invested 5.93 million US dollars to buy HYPE tokens. Account "0x55" spent 2.31 million US dollars to buy 59,719 HYPE at a price of 38.68 US dollars; account "0xe6" invested 2.11 million US dollars to buy 53,645 HYPE at a price of 39.30 US dollars; account "0x26" spent 1.51 million US dollars to buy 37,160 HYPE at a price of 40.7 US dollars, and simultaneously opened HYPE (10 times leverage) and BTC (20 times leverage)...
00:13 2025-06-18
Ohio passes HB 116 "Bitcoin Rights" bill
According to Cointelegraph, Ohio's "Bitcoin Rights" bill, HB 116, which protects self-custody rights, guarantees mining and node operations, and creates a $200 tax exemption on capital gains from digital asset transactions, passed the House Technology and Innovation Committee 13-0.
00:07 2025-06-18
Coinbase Announces Launch of Spark (SPK)
Coinbase has officially announced that it will support the Spark (SPK) token on the Ethereum network (ERC-20), which will be marked as an "experimental asset". Coinbase said that SPK transactions will be launched in phases once liquidity conditions are met.
00:04 2025-06-18
pump.fun and its founder's official X account has now been lifted
According to the X platform information, the official X account of the previously banned pump.fun and its founder Alon has now been lifted. Yesterday's news, X platform freeze tide hit, pump.fun and its founder ALON's official X account was frozen.
23:54 2025-06-17
WLFI: USD1 for the first time as an exclusive stablecoin to help Sahara AI complete its IDO overshoot
WLFI officials said that Sahara AI successfully completed the $8.50 million initial coin offering (IDO) of SAHARA tokens on the token issuance platform Buidlpad. The cumulative subscription amount of the event exceeded 74 million US dollars, exceeding the subscription target by 777%. USD1 contributed 79% of the total subscription ($59 million) and BNB contributed 21% ($16 million).
23:45 2025-06-17
Yesterday ARKB net outflow 191.40 million USD
According to Farside Investors monitoring data, yesterday ARKB net outflow of 191.40 million US dollars.
23:30 2025-06-17
Golden Morning News | List of important developments overnight on June 18
21:00-7:00 Keywords: xAI, JPMD, JD.com, GENIUS Act 1. Trump calls for Iran to surrender unconditionally. 2.xAI is reported to be in talks to raise $4.30 billion through equity financing; 3. JPMorgan Chase will issue a deposit token JPMD on the Base network on a pilot basis. 4. The SEC delayed approval of the Franklin XRP and Solana spot ETFs. 5. Coinbase seeks SEC approval to offer blockchain-based shares. 6. Richard Liu: JD.com will apply for stablecoin licenses in major currency countries arou...
23:03 2025-06-17
Israeli military spokesperson issues evacuation warning to designated areas in Iranian capital Tehran
An Israeli military spokesperson issued an evacuation warning to designated areas in the Iranian capital Tehran.
22:54 2025-06-17
The Federal Reserve will hold a meeting to discuss easing bank leverage requirements
The US Federal Reserve is to meet to discuss plans to ease leverage requirements for big banks, in what is expected to be the start of a broad programme to rethink bank rules. The Fed announced that it would hold a board meeting on June 25 to discuss changes to the so-called "supplemental leverage ratio", which requires banks to hold assets regardless of risk.
22:51 2025-06-17
Iran has prepared missiles and other military equipment to hit U.S. bases in the event of a U.S. war
According to the New York Times, Iran has prepared missiles and other military equipment to hit U.S. bases in the event of a U.S. war.