Gym-Anything-RT Kicks Off CUA Week: Frontier VLM Agents Score Only 5-9% in Real-Time Environments
wellecks · x · 2026-10-29
wellecks launches CUA Week (computer-using agents), with day 1 being Pranjal Aggarwal's Gym-Anything-RT: real-time environments for vision-language agents.
- Motivation: existing agent setups handle tasks where the agent can pause and think, but real-time acting agents lack training environments.
- Approach: Gym-Anything-RT synthesizes clever minigames that change as the model is thinking.
- Results: the environments are hard even for frontier models — GPT Astra scores 9%, Opus 5.5 only 5%.
- More projects will be released throughout CUA Week and aggregated on its page.
Related event: CMU L3 Lab launches CUA Week with a wave of agent research(3 posts)→
More from coding & agent
- Agents push compute back to CPUs: $221B server CPU market seen by 2030 — BenBajarin · 2026-10-29
- Cursor removes its beloved Cmd+K command, frustrating power users — jbhuang0604 · 2026-10-29
- Designing a skill like an app: a developer's iteration diary for interface-bts — round · 2026-10-29
- One prompt one-shots a walkable 3D ship where AI agents visibly work around the clock — InternationalGap3698 · 2026-10-29
- Dev suggests Codex, Claude Code, Cursor add a GitHub-style activity wall to flex AI usage — gethackteam · 2026-10-29
- mcp-coatcheck adds MCP Tasks to any server with one npx command, ending long-call timeouts — Awacer · 2026-10-29