Gemini 3.6 Flash
The structure specialist — not the smartest model, but the clearest thinking on the page.
Previous generation. Use Gemini 3.7 Flash instead. Superseded on 13 Aug 2026, 23 days after it shipped. Same price as 3.7 Flash, so there's no reason to stay on it.
- Maker
- Context
- 1M tokens
- Pricing
- $0.75 / 1M tokens / $3.75 / 1M tokens
Same introductory pricing as 3.7 Flash, through 31 Dec 2026. Google claims up to 17% fewer tokens than 3.5 Flash on the same work.
Best for
- Pressure-testing the narrative of a deck or doc you built elsewhere
- Imposing a clean structure on messy notes and research
- Quick analysis across many documents, at a working price
Prompting quirks
Bring it a finished artefact, not a blank page
Its edge is organising what already exists, not inventing what doesn't. Paste the deck, the doc or the argument you built in another model and ask whether the narrative holds and where the structure breaks.
Keep instructions tight and goal-first
Flash is optimised for speed; a crisp, unambiguous ask gets a crisp answer. Save the open-ended exploration for a flagship.
With the Pro line stalled, this is Google's real answer
Because 3.5 Pro hasn't shipped, Flash is where Google's current improvements actually land. Benchmark it against the Pro tier you're used to before assuming you need to wait.
Copy-paste example
Tuned for Gemini 3.6 Flash. Swap in your own content where marked.
{paste the deck outline, doc or argument you built elsewhere}
Don't rewrite this. Tell me whether the narrative holds: what's the
through-line, where does the logic jump, which section is in the wrong
place, and what would a reader still be asking at the end? Then give me
the restructured running order, section titles only.Facts verified against the official Google docs on 19 August 2026.