On July 21, 2026, Google released Gemini 3.6 Flash — not the long-awaited 3.5 Pro, but a leaner successor to 3.5 Flash that ships two things: 17% fewer output tokens on the same tasks, and a lower price.
Key specs
| Gemini 3.6 Flash | Gemini 3.5 Flash | |
|---|---|---|
| Input price | $1.50 / 1M | $2 / 1M |
| Output price | $7.50 / 1M | $10 / 1M |
| Context | 1M tokens | 1M tokens |
| Output efficiency | 17% fewer tokens | baseline |
What changed
- Price cut. $1.50/$7.50 is 25% cheaper than 3.5 Flash’s $2/$10 — and the 17% token reduction compounds on top.
- Drop-in replacement. Google positions 3.6 Flash as the direct successor; 3.5 Flash enters maintenance.
- Still no 3.5 Pro. The gap in Google’s lineup remains: 3.1 Pro ($7/$21) handles the heavy work, but the expected 3.5 Pro that would bridge cost and capability has not shipped. Google has confirmed Gemini 4 pre-training has begun.
Why it matters
- The workhorse got cheaper and leaner. For high-volume pipelines where Flash is already the default, 3.6 is a bill reduction without a migration — same API, fewer tokens, lower rate.
- Google’s model cadence is unusual. Skipping 3.5 Pro and shipping a Flash efficiency upgrade instead signals that Google is prioritizing the Gemini 4 generation over filling the current lineup.
- Flash as margin play. With 17% fewer output tokens, Google is essentially selling the same quality at ~35% lower effective cost when you combine rate and efficiency — a competitive move against Claude Sonnet 5’s introductory pricing.
Full specs: Gemini 3.6 Flash model page. For landscape context, see the 2026 LLM guide.