6,126Anthropic's Claude Code Now Tests Apple's Foldable iPhone Duo Layouts AutomaticallyClaudeDevs·Development·2 hrs ago·
3,654Alibaba's Qwen2.5-Math-PRM-7B Catches Exactly Where Your Math AI Goes WrongQwen·Reasoning·8 hrs ago·Pro
3,636AtomicChat Strips LFM2.5 Refusals From 96% to 1% by Editing Later BlocksAlphaSignal·Llms·8 hrs ago·Pro
3,614HKU's AutoSwarm Lets One Prompt Command Six Coding Agents at OnceAlphaSignal·Agents·22 hrs ago·Pro
3,609OpenAI Gives Four Safety Labs Unprecedented Access to Its Model TrainingOpenAI·Post Training·8 hrs ago·
2,833Appear2Meaning Benchmark Exposes AI's Blind Spot on Museum ArtifactsAlphaSignal·Image·18 hrs ago·Pro
2,725LiquidAI's d1.2-Ultra Tops Decision Index With Near-Perfect Confidence CalibrationLiquid AI·Llms·8 hrs ago·
2,505DeepReinforce's GrandCode Beats Every Human in Three Live Coding ContestsAlphaSignal·Development·6 hrs ago·Pro
2,469Tencent Drops Hy4 Flash at $0.07 per Million Tokens for High-Volume AI WorkTencent Hy·Llms·17 hrs ago·