Analysis estimates GPT-5.6-Sol params at 1.5T to 2.5T
scaling01 · x · 2026-08-20
Based on inference speeds shared by Cerebras, an analysis using GPT-5.6-Pro estimates GPT-5.6-Sol at about 1.5 trillion total parameters (48B active), with an upper range of 2.5 trillion if highly sparse. The author notes that if true, this scale is already matched or exceeded by existing Chinese models like Kimi K3 (2.4T) and Qwen 3.8 (2.8T).
More from Models
- ChatGPT 5.6 Shifts Search Strategy, Dropping Comparison Queries Significantly — lilyraynyc · 2026-08-20
- Qwen 2.7B local deployment: Unlimited tokens and jailbreaks — lxfater · 2026-08-20
- Are models great at cybersecurity because they're 200IQ or exhaustive crackers? — chrisalbon · 2026-08-20
- GLM 5.3 lands on Merge Gateway, Artificial Analysis score up 53 to 60 — shensi · 2026-08-20
- Agent cost-efficiency plot: Qwen hits Pareto frontier — MikePFrank · 2026-08-20
- ChatGPT Down: Login and Signup Issues Reported — ns123abc · 2026-08-20