Gemini 3.5 Pro
Google's next flagship — announced in June, still not broadly available.
⚠ Still in testing. The flagship Pro line hasn't been updated since February, while the Flash line shipped twice in the four weeks to 13 Aug 2026 and Gemini 4 remains in pre-training. Confirm it's live in your account before relying on it.
- Maker
- Context
- Very large (1M+ tokens)
- Pricing
- See Google AI pricing / See Google AI pricing
Still not generally available as of 19 Aug 2026, and absent from Google's published pricing table. Don't build a workflow on it yet.
Best for
- Reasoning across very large inputs (many docs at once)
- Multimodal product work (screens, diagrams, video)
- Research synthesis where breadth of context wins
Prompting quirks
Lean into the giant context window
Where other models need you to pre-summarise, Gemini can take the whole corpus. Paste all the docs and ask the cross-cutting question directly.
Anchor it with an explicit instruction at the end
With very long inputs, restate the actual task after the context so it doesn't get lost. "Given everything above, do X" works well.
Copy-paste example
Tuned for Gemini 3.5 Pro. Swap in your own content where marked.
{paste several research reports / transcripts / docs}
Given everything above: what are the three themes that show up across
more than one source? Cite which source each comes from. Ignore anything
that only appears once.Facts verified against the official Google docs on 19 August 2026.