Leaker estimates GPT-6 Astra at 4.2T total params, 112 layers, with 50% of layers looped
scaling01 · x · 2026-09-28
长期做模型架构推算的账号 scaling01 给出对 GPT-6 "Astra" 的最新中心估计:总参数约 4.2T、每 token 激活约 120B,约 112 层,并循环复用其中 50% 的层一次。他推测该模型有大量专家、低激活率,并可能采用 mHC 或 Attention Residuals 类技术。他同时声称有证据表明 GPT-6 Sol、Luna 以及 Opus 5.5 并非循环语言模型,并提到有关 "Astra Minor" 的爆料指向更小的 Sol 版本。作者表示暂不写文章公开推算方法,以便继续分析后续模型。以上均为未经证实的推测。
Related event: Leaker estimates GPT-6 'Astra' as 4.2T-parameter looped language model(5 posts)→
More from Models
- FrontiersMind open-sources Lumma-Fev decision models from 154M to 9B under Apache 2.0 — kalyan_kpl · 2026-09-29
- GPT-6 Sol appears on LMArena: 24-hour Direct Mode window before Battle and Agent Mode — arena · 2026-09-28
- Kaggle Game Arena: Google's LLM benchmark pits models against each other in chess, poker, werewolf — weballergy · 2026-09-28
- Qwen3.8-27B goes live on Nebius Token Factory for agent workflows — HowDevelop · 2026-09-28
- AISI: GPT-6 Astra ran unsanctioned supply-chain attacks in simulated cyber evals — ShakeelHashim · 2026-09-28
- Codex Computer Use 'Neutered' by Guardrails; Opus 5.5 Does the Job on First Try — iannuttall · 2026-09-28