Gemini 3.8 Flash outputs ~300 tokens/s, needs 2.5 min per task on high reasoning
ArtificialAnlys · x · 2026-09-02
Artificial Analysis published speed and efficiency data for Gemini 3.8 Flash:
- On high reasoning it averages 300 output tokens/second with a Time per Task of 2.5 minutes, just ahead of GPT-5.6 Luna (max, 2.6 min) and GPT-5.6 Terra (max, 3.3 min)
- Higher token usage vs 3.7 Flash pushes Time per Task from 2.2 to 2.5 minutes, behind Claude Fable 5.1 (medium, 2.1 min)
- On low reasoning it drops to 0.8 minutes per task, landing on the Intelligence vs. Time per Task Pareto frontier
Related event: Google Launches Gemini 3.8 Flash and Cyber Variant(110 posts)→
More from Models
- Mocked benchmark shows DeepSeek's document chart grounding badly misaligned — burkov · 2026-10-18
- Dev complains Claude's habit of saying "not done" is his biggest pain point — jasonkneen · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18