Google launches Gemini 3.6 Flash and says Gemini 4 training has already begun
op7418 · x · 2026-07-22
Google says Gemini 3.6 Flash is now live, alongside a faster Gemini 3.5 Flash-Light variant.
- Google describes 3.6 Flash as its “smartest everyday model,” with better coding, reasoning, and tool calling.
- The company claims it uses up to 65% fewer output tokens on benchmarks like DeepSWE while improving quality and precision.
- A Google spokesperson also said Gemini 4 training has already started and described it as the largest training run so far.
The repost adds a skeptical take that Google’s release cadence still looks slower than rivals that ship every few weeks.
Related event: Google Releases Three New Gemini Models Focusing on Efficiency and Cost(118 posts)→
More from Models
- Google is filling out capability, cost, and security models for agent builders — eyishazyer · 2026-07-22
- Frontier LLMs all lost money in a 1.6-year synthetic trading test — Scobleizer · 2026-07-22
- Reddit user says a chat-template tweak can force Laguna-S-2.1 into longer reasoning — SnooPaintings8639 · 2026-07-22
- Anthropic–Qwen distillation debate turns into a licensing argument — nptacek · 2026-07-22
- AI model pickers now bundle context length, reasoning level and speed toggles — jonathan_wilke · 2026-07-22
- Gemini trails Meta models on a shared leaderboard, according to the chart — IndividualShift2873 · 2026-07-22