Exa 3.0’s search token consumption has decreased by 47%, matching Perplexity in independent evaluations.

Beating AI News: Ant Group has open-sourced its Ming-3.0 voice model series, spanning three core domains—speech synthesis, audio understanding, and speech recognition—with a total of five models: 2 TTS, 2 Audio, and 1 ASR, all released under the MIT license. Developers can combine these models to build functions like voice input, content analysis, and voice responses. The standout is the 500-million-parameter Ming-3.0-tts-0.5B. Users can use natural language to command the AI to adjust tone (e.g., sounding angry), emphasize specific words, or modify speech rhythm. It also supports voice cloning, custom tone design, and inserting 14 sound tags (such as soft laughs and sighs) at designated positions. In Artificial Analysis’ same-voice blind listening test, the model ranks first among open-source models with an Elo score of 1069. The audio understanding model, Ming-3.0-audio-flash, is far larger, with 126 billion total parameters and ~6.5 billion active parameters per activation. It can directly process meeting recordings to answer questions, summarize consensus, disagreements, and action items without generating a separate transcript first. Another model, Ming-3.0-asr-tiny, specializes in speech recognition, supporting input of names, place names, industry terms, and conversation context to aid dialect and rare word recognition. Across four speech recognition tests conducted by Ant Group, its average error rate is 4.41%.

According to market data from BIT (bit.com), U.S. crypto-related stocks edged lower in pre-market trading, with the following declines among individual names: SBET down 0.8%, BMNR down 1.67%, CRCL down 1.07%, MSTR down 2%, COIN down 1.31%, and ASST down 1.37%

Per Lookonchain monitoring, Drift Protocol has resumed activity after two months of inactivity, transferring ETH to Tornado Cash. Over the past two hours, the attacker has deposited 25,700 ETH (worth roughly $61.56 million) into Tornado Cash, and currently holds 81,864 ETH (valued at approximately $195 million).

Beating AI News Flash from Dongcha: ZEALER founder and former GREE executive Wang Zirong has unveiled VIVA, the first AI product his team is developing, focused on personal health management. The tool integrates three virtual roles – a dietitian, fitness coach, and life assistant – into one chat interface. Users can either snap a photo of their meals or directly state what they ate; the AI will calculate calories and log nutritional intake. The fitness coach customizes training plans based on the user’s schedule, guides workouts, and tracks progress. The life assistant supports integration with Apple Health, adjusting daily recommendations using sleep and activity data – for instance, if the user got insufficient sleep last night, training intensity will be appropriately reduced today. The three roles share conversation information, so users don’t need to switch between separate tools. The product has a backstory: In June this year, Wang showcased an AI health management prototype he built, disclosing he previously used AI to aid fat loss, dropping his weight from 86kg to 74kg. During development, he prioritized solving two pain points: the complexity of Chinese home cooking makes calorie estimation hard, and smart wearable data is scattered, requiring manual sorting for AI analysis. In August, Wang revealed his startup’s software would launch no later than September. Now, the team plans to host on-site VIVA experiences at Xiaohongshu’s "Ma?ana Tomorrow Park" in Haikou from October 23 to 25.

The Drift Protocol exploiter, who stole over $270M, is moving $ETH to #TornadoCash after 2 months of inactivity. Over the past 2 hours, the exploiter has deposited 25,700 $ETH ($61.56M) into #TornadoCash and still holds 81,864 $ETH ($195M).

According to South Korean media reports, Samsung Electronics' Foundry Business Division, which handles semiconductor contract manufacturing, has begun targeting new customers for its 5-nanometer (nm, one billionth of a meter) process, as its 4-nanometer process capacity is already fully operational. While contracts for its most advanced 2-nanometer process with global large tech firms have been delayed, the move aims to boost production utilization centered on its 4-nanometer and 5-nanometer processes—boasting relatively stable yields—to improve profitability. Per semiconductor industry sources on the 20th, Samsung Foundry’s 4-nanometer process capacity has effectively reached saturation amid a recent rise in customer orders. As such, Samsung is expanding its sales efforts to attract new clients for its 5-nanometer process. Analysts note that this push for 5-nanometer customers is also a response to the demand shift: with 4-nanometer capacity nearing saturation, the 5-nanometer process can be offered as an alternative to meet clients’ performance and cost requirements, while also lifting utilization of existing production facilities. Conversely, securing major customers for its 2-nanometer process—its next-generation flagship technology—remains a key challenge. Samsung is in talks with global large tech firms for semiconductor manufacturing contracts based on its 2-nanometer process, but some negotiations are progressing slower than initially anticipated. Advanced processes require not only yield verification but also comprehensive validation of energy efficiency, design intellectual property (IP), and stable mass production capabilities, making the timeline to finalize contracts and launch mass production considerably long.