Sooool open-sourced: GPT-6 Sol as a real-time policy, with parallel requests to beat latency
wzenus · x · 2026-10-17
New repo Sooool explores GPT-6 Sol playing real-time games as "agent as policy": see the current screen, decide, send an action, observe, repeat. The core challenge is that the world keeps changing while the model processes frames, so an action may be wrong by the time it arrives. The built research loop accounts for this delay: up to eight model requests run in parallel, and Sol keeps looking at new frames and choosing actions while earlier decisions are still in flight.
The repo ships portable SOL visual policy evaluation, empirical latency profiles and simulation, the latency-aware autoresearch loop, multiple game environments with controls, results and research evidence, plus a quick start, code guide and citation.
Related event: wzenus open-sources Sooool: GPT-6 Sol plays real-time games frame by frame(8 posts)→
More from coding & agent
- Victris Arena: a programming game where humans and AI agents compete on one leaderboard — tristanbob · 2026-10-17
- os-taxonomy on GitHub: open curriculum graph with 1,590 topics and 5k stars — aigclink · 2026-10-17
- Open-sourced os-taxonomy: 1,590 elementary knowledge points, 3,221 prerequisite edges — aigclink · 2026-10-17
- RSI-top framework unblocks stalled research agents, hits 77.3% medal rate on MLE-bench Lite — dbreunig · 2026-10-17
- Vibe coding wisdom: the line between AI psychosis and next-level is unit tests — broodsugar · 2026-10-17
- CoPE adds one loss to GRPO, boosting language agent sample efficiency up to 2× — Lianhuiq · 2026-10-17