Google released Gemini 3.7 Flash on 13 August with an introductory API price of $0.75 per million input tokens and $3.75 per million output tokens. The offer runs through 31 December 2026; the announced rates from 1 January 2027 are $1.50 and $7.50 respectively.
The release announcement positions the model for coding, document work and agents that use tools across several steps. It follows Gemini 3.6 Flash by three weeks.
The price has an expiry date
A team estimating the cost of a continuing service needs to distinguish the introductory rate from the announced longer-term rate. At the same token volume and input-output mix, the January prices would double the token charge relative to the introductory offer.
That arithmetic does not determine the total cost of an application. A workflow may also incur tool charges, storage, retrieval and review costs. The number of attempts needed to complete a task can matter as much as the price of an individual request.
For example, a lower token price would not automatically make a process cheaper if it repeatedly produced outputs requiring extensive repair. Conversely, a model that completes a task with fewer retries could reduce usage. Those are questions for testing on the application’s own work, not conclusions established by a launch price.
The performance evidence comes from Google
Google reports improvements over 3.6 Flash on several evaluations. Its published FrontierCode 1.1 Main scores are 43.6% for the new model and 34.4% for its predecessor. It also reports stronger results in document processing, business workflows and web development.
These are vendor-reported benchmark results. They indicate what Google measured under its evaluation conditions, but they do not establish the same improvement for every codebase, document set or agent configuration.
The announcement also says the model follows instructions more reliably and handles multi-step planning and tool calls more effectively. A practical evaluation should include failures, permission boundaries and recovery from incomplete actions, alongside successful outputs.
Spark receives the model too
Google says Gemini Spark begins using 3.7 Flash from the release date. Spark is available to Google AI Pro and Ultra subscribers in more than 160 countries, according to the announcement.
For organisations already considering AI inside document workflows, the model update adds another reason to keep evaluation records tied to a specific model and date. A test of an earlier Flash release is not automatically a test of this one. The temporary price should likewise remain explicit in any budget comparison.
Questions
When does Gemini 3.7 Flash’s introductory price end?
Google says the offer expires on 31 December 2026, with the announced higher rates applying from 1 January 2027.
What are the introductory token prices?
The release lists $0.75 per million input tokens and $3.75 per million output tokens.
Are the benchmark results independent tests by this publication?
No. The reported scores come from Google’s launch material and need testing against a team’s own workload.
