PulseAugur
EN
LIVE 15:10:04

Moonshot AI's "3T-class" model sparks debate over parameter count

Moonshot AI is reportedly marketing its new AI model as "3T-class," despite it having 2.8 trillion parameters. This has drawn attention for its unconventional approach to numerical classification. The model is also associated with the "Pelican Benchmark," a comparison that has been questioned for its relevance to AI capabilities. AI

IMPACT Raises questions about industry marketing practices and the validity of AI benchmarks.

RANK_REASON The item discusses commentary on a company's marketing claims regarding its AI model's parameter count and benchmark relevance.

Read on Mastodon — fosstodon.org →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Moonshot AI's "3T-class" model sparks debate over parameter count

COVERAGE [1]

  1. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    🚀 Apparently, calling a 2.8 trillion parameter # AI model a "3T-class" is the new # math lesson from # Moonshot AI! 🤔 But don't worry, there's still time to dis

    🚀 Apparently, calling a 2.8 trillion parameter # AI model a "3T-class" is the new # math lesson from # Moonshot AI! 🤔 But don't worry, there's still time to dissect the "pelican benchmark"—because comparing AI to birds is super relevant now. 🐦💡 https:// simonwillison.net/2026/Jul…