Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
Google has released Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, two variants that share the same underlying architecture but differ in their safety configurations. While Gemini 3.8 Flash delivers coding performance comparable to Claude Opus 5, its increased reasoning demands lead to a 30 percent higher consumption of output tokens per task. This shift suggests that despite steady per-token pricing, the model may cost more to operate in practice than previous versions, marking a new trade-off between agentic capability and computational efficiency for developers.
ModelsGemini 3.8 Flash
Covered by 6 sources · 7 articles
- GGoogle DeepMind Blog↗Sep 2
- TThe Decoder↗Matthias BastianSep 2
- TThe Decoder↗Jonathan KemperSep 2
- MMarkTechPost↗Asif RazzaqSep 2
- TThe Verge↗Stevie BonifieldSep 2
- AArs Technica↗Ryan WhitwamSep 2
- HHacker News↗brataoSep 2