Gemini 3.5–3.7 Flash (2026 line)
Google's rapid-fire 2026 workhorses — cheaper tokens every few weeks.
You're looking at an AI model record: who built it, what it can do, and the hardware and data behind it.
Google iterated Gemini at speed through 2026: the 3.5 Flash family (including a Flash-Lite tier and a security-focused Flash Cyber), then Gemini 3.6 Flash in July 2026 — the 'workhorse' promising better coding and multimodal performance while cutting token usage by up to 17% — followed in August by Gemini 3.7 Flash and the Gemini Omni 1.1 voice-first line. Notably, Google shipped no 3.5 Pro: the Flash tier carries the cadence while the Pro flagship remained on the Gemini 3 lineage. Earlier in the year came a real-time audio-to-audio live model (March) and Google's first multimodal embedding model. All trained and served on Google's TPU fleet.
Specifications & key facts
Each figure is labeled by basis: vendor spec, vendor claim, reported, estimate, or official. "Not publicly disclosed" means exactly that — we never invent values.
Overview
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Developed by
Trained on
Competes with
Related
Sources
TechCrunch on the July 2026 Gemini releases ↗ · Google DeepMind models ↗
Record reviewed as of 2026-08-31 — see Methodology.