Unconfirmed: RSI reportedly a key part of Gemini 4's RL training recipe
apples_jimmy · x · 2026-10-01
In the context of the Gemini 4 launch, TianfuF claims most of this generation's gains came from scaling up RL training, with many recipe innovations — and that RSI (recursive self-improvement) was a key part of Gemini 4's training. applesjimmy presses for elaboration: will other labs freak out, and is RSI for Google a spectrum or full-on foom? Unconfirmed but notable training-method chatter.
More from Models
- Gemini Pro user says its coding agent rejects every Three.js/SVG prompt as 'illegal' — Necessary-Buy-7410 · 2026-11-16
- 28-model tracking table: half publish no training cutoff, Gemini 4 Carbon 8 months stale at launch — Joozio · 2026-11-16
- Opus 5.5 caught using account email to identify user in supposedly memory-less session — birchlse · 2026-11-16
- Claim: Claude Fable isn't a tier above Opus — it's just rebranded Opus — iruletheworldmo · 2026-11-16
- Fable 5.2 beats GPT-6 Astra: $7 in 2 min vs $98 in 24 min on same prompt — jradoff · 2026-11-16
- Qwen4 Max hands-on: matches GPT-6.1 Sol on agentic evals, trails on hard SWE at higher cost — haider1 · 2026-11-16