Google introduced Gemini 3.7 Flash today. The company calls it the most intelligent version yet in its Flash line of models built for coding and automated agents.
The release comes just three weeks after Gemini 3.6 Flash. Google said the update reflects feedback from developers along with new technical improvements.
Gemini 3.7 Flash is priced at half the cost of its predecessor per million tokens. The new price is set through the end of the year.
Coding and Web Development Gains
The model performs better on coding tasks like debugging and fixing software issues. It also produces more accurate code on the first try.
On the FrontierCode 1.1 Main test, Gemini 3.7 Flash scored 43.6 percent compared to 34.4 percent for 3.6 Flash. On DeepSWE v1.1, it scored 65.3 percent versus 49.0 percent.
For web development, the model builds more complete and working web pages using fewer prompts. It can also match a design shown to it, whether that design is a screenshot or a full style guide.
On WebDev Arena, a testing platform for web development models, Gemini 3.7 Flash earned an Elo score of 1588. Gemini 3.6 Flash scored 1538 on the same test.
The model also improved in fields that require dense reading and reasoning, such as finance, law, and biosciences. On the GDP.pdf benchmark, which tests document comprehension, it scored 34.0 percent compared to 22.0 percent for the older model.
On AutomationBench, a test for completing real business tasks, Gemini 3.7 Flash scored 30.4 percent. Gemini 3.6 Flash scored 17.0 percent on the same benchmark.
Google said the model plans multi-step tasks with more care. It also needs fewer retries and less manual oversight during coding work.
Gemini Spark and Where to Access It
Gemini Spark, Google's personal AI agent, will run on Gemini 3.7 Flash starting today. Spark is available to Google AI Pro and Ultra subscribers in more than 160 countries.
With the update, Spark can complete Workspace tasks such as organizing files, writing emails, and updating status documents. Google said the tool works with less user input needed to finish these tasks.
Developers can access Gemini 3.7 Flash through Google Antigravity, the Gemini API in Google AI Studio, and Android Studio. Businesses can reach the model through the Gemini Enterprise Agent Platform and the Gemini Enterprise app.
Google also updated safety protections tied to the new model. These cover risks in chemical, biological, radiological, and nuclear areas, along with cyber offense risks.
The company said these safeguards follow its existing approach to bioresilience and cyber security programs. Full details are listed in the Gemini 3.7 Flash model card, which Google published alongside the release.
Gemini 3.7 Flash is now live across Google's developer, enterprise, and consumer products as of today.
