OpenAI unveils first custom inference chip Jalapeño with industry-leading throughput over Blackwell
Designed specifically for LLM inference and taped out in 16 months, benchmarks show superior efficiency per kilowatt over current market systems.
ByteDance launches Doubao Work desktop agent, integrating Feishu, Coze, and developer tools
Unifies workplace context to execute cross-application desktop workflows with persistent cloud execution.
Inside this issue: 3 top stories · 2 research papers · 3 tools and products · 3 builder perspectives · 3 community stories · 2 trending repositories
Read the full issue →Vermilion marks published dates; the outline marks today.