THIS APPLE PAPER IS F*CKING INCREDIBLE
This is a dev post classified by AGK Intelligence as AI dev tools (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
THIS APPLE PAPER IS F*CKING INCREDIBLE Apple just proved the visual agent harness thesis from my post below Instead of burning test-time compute on blind internal reasoning, their agent harness beats Codex and Visual Sketchpad while cutting token cost by 63%: most developers trying to build visual agents assume that scaling inference compute or test-time thinking solves real-world perception. Apple evaluated this across 22 benchmarks and proved that scaling internal thinking alone yields diminishing returns while burning thousands of dollars in model tokens Apple researchers analyzed multi
Posted by marfin (5.6k followers) 3 h ago · 36 likes · 1.8k views · view the original post on X. Kept by the Dev Radar as AI dev tools. Tools mentioned: openvlharness.github.io.
More dev work like this
- 🎨 Create Images with GitHub Copilot in #VSCode #GitHubCopilot — @code
- New for Enterprise: Custom Providers API — @raycast
- instantly added this, go go @ChatGPT Dot as my video editor in the cloud — @theaaron
- GLM 5.3 is live on Chutes. — @chutes_ai
- I tested LongCat-2.5 and Qwen3.8-Max on the same browser game, and LongCat got it done… — @omarsar0
- New in Superconductor this week: — @Superconductor
- @MiniMax_AI open-sourced BYOB Table Protocol, a model-agnostic protocol where you bring… — @javierbits
- Another great visibility hack for your SaaS is to get your toolkit on Composio (DR 77).… — @tarasshyn
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.4k posts from 5k X accounts over the last 21 days, 9.1k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-03 15:12 UTC. Full method.