TokenRhythm unveils NeoHorse-1: 4B/9B models claiming first closed-loop step toward RSI
jiqizhixin · x · 2026-09-19
TokenRhythm, with algorithm research from Tsinghua and Peking University and infrastructure from Wwenxin (Infinigence), released NeoHorse-1 (4B/9B parameters), billed as the first closed-loop verification toward Recursive Self-Improvement (RSI).
- Problem: agent execution experience stays in logs and rarely enters model weights, so models repeat the same mistakes on similar tasks
- Method: the routing system that dispatches tasks also records every execution path models take; these trajectories are distilled into a small model that can independently reproduce what a whole fleet of models collectively learned
- Framed as an "agent experience into parameters" training paradigm and a first step toward RSI
More from Models
- Meta's Muse personal agent spotted working in Canada, user posts hands-on note — hwbhatti · 2026-09-19
- Astra is the fastest-growing expensive model, enterprises only ask for lower price — bindureddy · 2026-09-19
- Jev is first AI lab to officially release a hallucination rate benchmark — StewartalsopIII · 2026-09-19
- Grok Voice Transcribe 2.0 Reported: 2x Accuracy, #1 Among 32 Streaming STT Models (Unverified) — NicoVerderosa · 2026-09-19
- Rumored Gemini 4 Pro pricing: $3/M input, $12/M output, 5x cheaper — bindureddy · 2026-09-19
- GPT-6 Astra Wins Only 2 of 10 Livestreamed Slay the Spire 2 Runs at Ascension 10 — Jsevillamol · 2026-09-19