Gemini 3.8 Flash outputs 48k tokens per task on high reasoning, 30% more than predecessor

ArtificialAnlys · x · 2026-09-02

Artificial Analysis notes Gemini 3.8 Flash averages 48k output tokens per task on high reasoning, up 30% from 3.7 Flash, at 300 tok/s; Time per Task rises from 2.2 to 2.5 minutes, behind Claude Fable 5.1 (2.1 min). Low reasoning outputs 14k tokens per task in 0.8 minutes.

Related event: Gemini 3.8 Flash Benchmarks: Smarter, Pricier, Still on the Cost Frontier(8 posts)→

Original post →

More from Models

Models channel →