Lookonchain APP

App Store

Hermes Founder to Launch AI Game Studio, Intends to Delegate Coding, Art Development, and Even Playtesting and Bug Hunting to AI Agents

1 hours ago

Beating AI Express News: Nous Research co-founder Teknium announced plans to rework Hermes into a DIY game development studio. Developer Luke proposed a multi-sub-Agent collaborative workflow: one Agent handles game logic coding, another creates character and scene assets, a third is responsible for music and sound effects, and a dedicated Agent playtests the game, reports bugs, and even identifies unengaging areas. Teknium then outlined required capabilities to add, including 3D model generation, animation, sound effects, and integration with Unity and Godot game engines. He clarified all features will be implemented via plugins, with no modifications to Hermes’ core. Luke also suggested adding NPC voice acting, 2D pixel assets, and one-click export for web and mobile game versions. Hermes already has built-in programming, image generation, and sub-Agent collaboration capabilities. The team previously launched a Blender plugin and supports Unreal Engine editor control via MCP, with pre-existing options for music and sound effect generation. Similar concepts have emerged before: open-source project Claude Code Game Studios has been leveraging specialized multi-Agent collaboration for game design, development, and testing since February this year, and community developers later adapted it into a Hermes version.

Relevant content

OpenAI 28-Day Sprint Day 15: Tibo goes from "We're doomed" to "Still working on it", uses quota reset to fend off the day

Beating AI News Flash: OpenAI’s 28-day product sprint is now on Day 15. Tibo first shared a ramen-cooking show titled "COOKED" with the caption "We're cooked" (literally meaning "we’re done for"). About 11 hours later, Tibo backtracked, posting: "Never mind, we’re still working on it… here’s another quota reset, enjoy." This round of reset has now taken effect.

2 minutes ago

Goldman Sachs is bullish on ChangXin against the trend: Q3 revenue beat expectations by 30%, and mass production of 5th-generation DRAM has fully unlocked growth potential.

Goldman Sachs released a new research report, maintaining its "Buy" rating for ChangXin Memory Technologies and slightly raising its 12-month target price from RMB 129 to RMB 130. The direct catalyst for the hike is ChangXin's disclosed third-quarter earnings preview: Goldman Sachs projects the company's Q3 revenue will range between RMB 115 billion and RMB 120 billion, marking a 15% to 20% quarter-over-quarter increase, with the midpoint roughly 30% higher than Goldman's prior forecast. This better-than-expected guidance, coupled with the company's announcement that its 5th-generation DRAM technology platform (G5) will enter mass production in September 2026, further reinforced Goldman Sachs analysts' confidence in ChangXin's medium- and long-term growth thesis. Accordingly, Goldman raised the company's profit forecasts for 2026 to 2030, and labeled it a core beneficiary amid China's DRAM demand upcycle. The company's current share price stands at around RMB 40, implying an upside potential of as high as 225% based on Goldman's target price. Notably, ChangXin's share has fallen sharply recently, with its current market capitalization dropping below RMB 3 trillion.

2 minutes ago

Analyst: The market is at a critical juncture of a structural correction, with the futures market still dominated by sellers.

Crypto market analyst Axel Adler Jr. released his latest market observation, noting that Bitcoin has fallen around 6% from $86,700 on October 2 to $81,600 on October 20. The market is currently at a critical juncture for a structural correction, with bulls and bears battling near key support levels. According to Alex’s analysis, Bitcoin’s market structure signal has dropped from +0.37 a week ago to -0.32. Although the signal has moved out of the bullish range, it remains in the neutral zone and has not yet hit the bearish threshold below -0.5. He emphasized that the lower bound of the price range over the past three weeks is $80,400. As long as prices hold above this level, the current movement can still be considered a normal correction within the range; however, if prices break below $80,400 and the structure signal simultaneously drops below -0.5, the market correction will officially evolve into a downtrend. In the futures market, the index line reflecting the strength of long and short positions (buyers above 50, sellers below 50) has been below 50 since October 17. Although the index approached 50 on October 19, it failed to break through, and has now fallen back to 46. Alex stated that a decisive break above 50 for the index line will be the first signal of buyers returning. However, sellers currently hold absolute dominance, and futures market demand lacks sufficient momentum to push prices higher. Overall, Bitcoin is currently in a critical period of direction determination. The market needs to closely monitor the performance of the $80,400 support level and whether the futures long-short index can reclaim the 50 threshold, to determine whether this correction is a short-term adjustment or a precursor to a trend reversal.

2 minutes ago

DeepSeek teams up with Huawei to challenge NVIDIA's AI computing hegemony: participating in next-generation chip design and pushing ahead with pre-training of domestic chips by the end of the year.

Beating AI News (Dongcha) reports that last year, DeepSeek attempted to train its R2 model using Huawei Ascend chips but failed. The Financial Times previously disclosed issues with chip stability, inter-chip communication, and software support. Even with Huawei engineers on-site assisting, DeepSeek ultimately switched back to Nvidia chips to complete the training. Over a year later, the two parties have deepened their collaboration to include hardware design. The Information now reveals that Huawei engineers are permanently based at DeepSeek’s offices, integrating the practical demands of model training directly into the design of next-generation Ascend chips—including how to schedule computing tasks and transfer data between chips. At the end of September this year, DeepSeek open-sourced the Ascend version of TileLang, plus underlying tools for matrix computation, MoE expert communication, etc., to reduce the development work required for migrating models to domestic chips. However, getting the software to run is still a long way from full pre-training. Tens of thousands of chips need to operate continuously for weeks; any single chip failure or data transmission delay could slow down the entire cluster. Now, the two sides are starting to tackle these issues at the chip design stage. Huawei plans to prioritize allocating most of its supply of Ascend 960DT, set to be ready in Q1 2027, to DeepSeek rather than distributing it to other model companies. DeepSeek, for its part, is preparing to start pre-training as early as the end of this year using the 950DT. So far, related tests have used over 1,000 chips, but full pre-training may require tens of thousands, and some cluster management software remains incomplete. Meanwhile, DeepSeek is still seeking more Nvidia chips.

2 minutes ago

OKX to List xNFLX USDC Spot Trading

According to an official announcement, OKX will launch xNFLX USDC spot trading at 19:00 Beijing time on October 20.

2 minutes ago

South Korean stock market closed down nearly 4%, SK Hynix dropped 4.22%.

According to Bitget market data, South Korea’s KOSPI index closed down 214.93 points on Tuesday, October 20, with a 3.73% decline, ending at 5553.08 points. Samsung Electronics fell 2.98%, while SK Hynix dropped 4.22%.

2 minutes ago

Popular tokens

BitcoinEthereumHyperliquidSolanaTRONBNBTetherAaveXRPPepeFartcoinOndoJupiterUniswapBonkPendleEthenaArbitrumAvalancheLidoChainlinkPolygonDogecoinCardano