Google’s Gemini 3.7 Flash Just Dropped 50%: Why the 2027 Price Cut Is a Rare Window
Google has given developers a compelling reason to test Gemini 3.7 Flash now: its API currently costs roughly half what the model will cost from January 2027.
Google released Gemini 3.7 Flash on August 13, just three weeks after Gemini 3.6 Flash. The new model is aimed at coding, debugging and agent workflows, with introductory pricing of $0.75 per million input tokens and $3.75 per million output tokens. That pricing lasts through December 31. On January 1, 2027, both rates double.
Price Doubles In January
Google is not hiding the increase. Its official documentation and pricing page state that the introductory rates expire on December 31, 2026, with standard pricing beginning the following day.
| Usage | Until Dec. 31, 2026 | From Jan. 1, 2027 |
|---|---|---|
| Input tokens | $0.75 / 1M | $1.50 / 1M |
| Output tokens | $3.75 / 1M | $7.50 / 1M |
| Cached input | $0.075 / 1M | $0.15 / 1M |
For a request using 100,000 input tokens and 10,000 output tokens, the token cost works out to about $0.1125 during the introductory period and $0.225afterward.
At larger volumes, the difference becomes harder to ignore. One hundred million input tokens cost $75 today and $150 from January. One hundred million output tokens rise from $375 to $750.
Model Targets Coding And Agents
Google is positioning it specifically for coding, debugging and agent workflows. Its published benchmarks show FrontierCode 1.1 Main improving from 34.4% on Gemini 3.6 Flash to 43.6%, while DeepSWE v1.1 rises from about 49% to 65.3%.
“Gemini 3.7 Flash is our most intelligent workhorse model yet for coding and agents,” said Tulsee Doshi, Senior Director of Product Management, writing on behalf of the Gemini team in Google’s August 13 announcement.
“Gemini 3.7 Flash supports a 1M token context window, 64k max output tokens, tunable thinking levels (low, medium, high), and the same suite of built-in tools as 3.6 Flash,” Google says in its developer documentation.
Those features make it particularly relevant to software agents that need to work through multiple steps rather than answer one question and stop.
Agent Workloads Make Costs Compound
A normal chatbot exchange might consume a relatively predictable amount of input and output. An AI coding agent is different.
It can send a request, call a tool, inspect the result, revise its approach, retry a failed operation and continue until the task is complete. Every one of those steps consumes tokens.
That means the difference between $0.75 and $1.50 per million input tokens, or $3.75 and $7.50 per million output tokens, can become significant when an agent is running continuously across thousands of tasks.
“From January 1, 2027, standard pricing will take effect,” Google says in its Gemini API documentation, which identifies the December 31 expiration date for the introductory rates.
Caching Can Keep Costs Lower
Gemini 3.7 Flash’s cached input is priced at $0.075 per million tokens during the introductory period, compared with $0.75 for regular input. That is a 90% reduction.
The cached rate rises to $0.15 per million tokens when standard pricing begins, but the discount remains substantial.
Google lists Gemini 3.7 Flash context caching at $0.075 per million tokens through December 31, 2026, rising to $0.15 per million tokens from January 1, 2027. Instead of paying the full input rate every time, developers can reuse cached material and reduce the cost of repeated requests.
Budget For The Permanent Price
Google’s approach isn’t unusual for cloud infrastructure. A provider can use an introductory price to encourage developers to test a new model before moving toward its standard economics. The important part is that Google has given developers a specific date.
That makes this less of a pricing surprise and more of an infrastructure-planning issue. Developers testing Gemini 3.7 Flash have roughly four and a half months to evaluate the model at half its eventual input and output rates. But production budgets should not be built around those temporary numbers.Real Takeaway For Developers
Gemini 3.7 Flash’s introductory pricing is a genuine opportunity for developers building coding and agent workloads, particularly while the model’s performance can be evaluated at half its eventual token cost.
