Estimating Frontier Model Sizes: Claude Fable at 4.5T Parameters
anpaure · x · 2026-08-03
The author estimates the implicit total parameters of frontier models using intelligence indices, a No-CoT reasoning benchmark, API pricing, compute trends, and a poll of 20 AI researchers.
Results indicate Claude Fable 5 is around 4.5T, GPT-5.6 Sol is 3.1T, Claude Opus 5 is 3.0T, and Kimi K3 is 2.8T. The methodology aims to make opaque parameter counts inspectable and reproducible.
More from Models
- GPT-5.6 Luna Max in Codex Outperforms Sol at Fraction of the Cost — DeryaTR_ · 2026-08-03
- Experiments Show Existing Public LLMs Can Recreate Project Astra Results — danshipper · 2026-08-03
- MiniMax H3 Encoder Leaks; Devs Debate Quantization vs Prompt Adherence — ayakitodev · 2026-08-03
- Comparing Math Proofs: GPT-4o vs. Astra — danshipper · 2026-08-03
- Rude Prompts Cut Token Usage by 44% in LLMs, Penn Study Finds — rohanpaul_ai · 2026-08-03
- ChatGPT Fails at PC Building Guide: Free Tier Lobotomized and Disconnected from Average Users — Kyokyodoka · 2026-08-03