Binde Reddy Calls Frontier Launches Disappointing, Predicts Open-Source Gap Hits Zero by Year-End

petergyang · x · 2026-09-23

Investor Binde Reddy argues recent frontier model launches "don't make any sense" after his own testing: GPT 6 Sol is cheaper but "way more terrible," while Opus 5.5 sounds more normal but shows no leap in intelligence. His thesis: if frontier labs stall, open source catches up faster and the gap drops to zero by year-end. petergyang replies questioning whether it's even worse than 5.6 Sol. Model version names are unverified; conclusions are one tester's subjective take.

Related event: Founder tests frontier models, finds them disappointing and predicts open-source parity by year-end(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →