APK Arena: New Mobile Agent Benchmark Released
giffmana · x · 2026-07-12
发布了一个新的前沿基准 APK Arena，用于评估 AI agent 在手机上的操作能力。
作者称他们对 GPT 5.6 做了横向对比，结果它在这套基准上整体领先于其他模型。基准覆盖 50+ 关卡，包括点击、视觉理解、记忆、游戏和真实任务等手机使用场景。
More from Apps
- Substack adds an in-app AI detection tool through Pangram integration — moultano · 2026-07-22
- AI agent designers map the visual and tonal cues behind companionship products — Unlikely-Platform-47 · 2026-07-22
- Strangeworks launches Aura to turn enterprise ops into production optimization systems — whurley · 2026-07-22
- OpenAI rolls out voice in GPT-Live, but the UI obscures search and reasoning — Graham_dePenros · 2026-07-22
- Reddit asks why online stores still haven't exposed product catalogs through MCP — gelembjuk · 2026-07-22
- Meta AI adds interleaved image-and-text input to its text box — ezyang · 2026-07-22