WHAT NEW IN X10.29 THU

AI collections from X · 22 collections · 58,925 posts

gemini-4POST

Gemini 4 Argon data analysis test claims 15% hallucination rate and top Vals Index

StatsWire says Gemini 4 Argon beats GPT-6 Astra and Claude on heavy data analysis, citing a 15% hallucination rate, 68.9% on the Vals Index, 65.4% on Finance Agent and a 1M context window.

Stats Wire
@StatsWireArchived 12 days agoOriginally posted 2026.10.15
Stats Wire @StatsWireOriginally posted 2026.10.15
Open original on X

I’ve been testing Gemini 4 Argon on heavy data analysis, and it’s easily Google’s strongest enterprise model yet. Why it beats GPT-6 Astra and Claude for serious data work: 1️⃣ 15% Hallucination Rate → 3.4× lower than Astra (51%) and Sol (54%) on AA-Omniscience 2️⃣ #1 on Vals Index (68.9%) & #1 on Finance Agent (65.4%) 3️⃣ 1M Context / 64K Output → Ingests massive CSVs and multi-file reports in 1 pass without dropping rows 4️⃣ Native Workspace Memory → Connects directly across Google Sheets, Drive & Docs at $1.99/task Coding gets all the hype, but data analysis is where Argon quietly took the crown.

x.com/statswire/status/2110578103890416035Archived 2026.10.16 10:14