Emad Mostaque: Luna reasoning surpasses GPT-5, top-tier AI will be free in 2 years
jdjohnson · x · 2026-08-07
Emad Mostaque shared his thoughts on OpenAI's latest model matrix. He noted that Luna in non-reasoning mode is slightly better than the SOTA model from two years ago, GPT-4o. Furthermore, Luna with medium thinking outperforms last year's SOTA, GPT-5 (High mode).
He emphasized that these powerful capabilities are now available for free and unlimited to everyone. Based on this trend, he predicts that within two years, AI at the Sol Max / Fable Max level will also be completely free and significantly faster.
More from AGI Musings
- Founder Mocks AI Innovation: Just Coding with Agents Months Before Big Tech Does — nptacek · 2026-08-07
- Software Veteran Masters Biology in a Year Using AI as 24/7 Tutor — JosephJacks_ · 2026-08-07
- Pessimistic and Optimistic Futures of Biosecurity: A 13k-Word Deep Dive — owl_posting · 2026-08-07
- Simon Willison's Top Tip for Tech Blogging: Lower Your Standards and Publish — rseroter · 2026-08-07
- AI Coding Tools Break Language Barriers, Shifting Focus to Client Communication — Vjeux · 2026-08-07
- The AGI Social Contract: From Fulfilling Work to Zero-Sum Bickering — danfaggella · 2026-08-07