Pricing and budget · 2026-09-22

Gemini 3.7 Flash: introductory pricing ends 1 January 2027, preparing your budget

Covers that Gemini 3.7 Flash's introductory pricing, valid through 31 December 2026, switches to standard rates (double the input/output price) on 1 January 2027, and how to prepare for that cutover.

Timeline showing Gemini 3.7 Flash at introductory pricing through 31 December 2026 and at double the standard price from 1 January 2027 onward.

The price exactly doubles, not gradually

Gemini 3.7 Flash's pricing card splits into two periods: through 31 December 2026 at introductory pricing, input text is $0.75 per million tokens and output is $3.75 (image, video, and audio input share the same $0.75 unit price). From 1 January 2027 onward, these figures exactly double: input rises to $1.50, output to $7.50. This isn't a gradual increase — it's a clean threshold that kicks in on a single date.

Prompt cache prices change at the same ratio: cache read goes from $0.075 in the introductory period to $0.15 in the standard period, and cache storage (per hour) goes from $0.50 to $1.

How a production budget should prepare for this cutover

If you're currently planning a budget at introductory pricing, separately calculating the post-1-January-2027 cost and presenting both scenarios to management or the budget owner prevents a surprise bill increase on that date. In a high-volume production flow, this 2x difference can be a meaningful line item in total operating cost.

Google Search and Google Maps grounding tools also carry a separate charge ($14 per 1,000 calls, the same in both periods); don't forget to add this tool-call cost alongside token pricing when calculating total cost for a flow using these tools.

  • Introductory pricing: through 31 December 2026, input $0.75 / output $3.75.
  • Standard pricing: from 1 January 2027, input $1.50 / output $7.50 (exactly double).
  • Grounding tools (Search, Maps) are separately priced and unaffected by the price change.

Reasoning defaults to the medium level

The model carries an adjustable thinking depth across low, medium, and high; the default is medium. Dropping to low on simple tasks reduces latency and cost, while raising to high on complex coding and analysis tasks provides deeper evaluation. This choice applies regardless of the introductory/standard price cutover and forms a second dimension of budget optimization.

Frequently asked questions

Does the price change gradually on 1 January 2027, or all at once?

It's a clean threshold kicking in on a single date; introductory pricing applies through 31 December 2026, and standard pricing applies from 1 January 2027 onward.

Does the grounding tools' price also change in this cutover?

No, Google Search and Google Maps grounding calls ($14 per 1,000 calls) are the same price in both periods; only the token-based input/output/cache prices change.

Related posts