Accusations fly that Artificial Analysis rankings systematically underrate open-weight models
aiamblichus · x · 2026-10-28
X users accuse eval outfit Artificial Analysis of deliberately scoring open-weight models lower, questioning how Gemini 4 Argon could beat DeepSeek V4.1 on agentic work. One calls it 'blatant manipulation' by an AI industry panicking about open source—unverified, but a notable trust crisis for third-party evals.
More from Fun
- Beff Jezos raves about Astra Ultrafast's token speed — beffjezos · 2026-10-28
- AI CEO memes about girlfriends leaving over Dyson swarm ambitions — beffjezos · 2026-10-28
- Users notice ChatGPT almost always ends replies with "One question:" — Angaisb_ · 2026-10-28
- jobergum's RAG-variant meme list pokes fun at the hype: vanillaRAG, grepRAG, DecisionRAG and more — jobergum · 2026-10-28
- AI-assisted porting craze gets wild: Xiaomi 15 boots iOS 16 — op7418 · 2026-10-28
- Manus pays a user to run her LEGO custom-set startup, first order in 2 days — henloitsjoyce · 2026-10-28