Google Says Gemini 3.6 Flash Cuts Tokens, Sharpens Coding Scores
The output-token cut reaches 17% against 3.5 Flash, with gains in knowledge work and multimodal tasks across Google's three-model Flash lineup.
- Gemini 3.6 Flash cuts output tokens up to 17% versus 3.5 Flash, Google says.
- Google reports gains in coding, knowledge work and multimodal tasks over the prior Flash model.
- Gemini 3.5 Flash-Lite hits 350 tokens per second; 3.5 Flash Cyber stays limited to governments.
Why it matters: Nobody outside Google can verify whether Gemini 3.6 Flash truly costs less to run.
Google Blog (Gemini team) ↗ · Jul 21, 20267/21/26 · Updated Jul 21 · ✓ Checked✓ Check