Google has released Gemini 3.7 Flash, a new AI model aimed at coding, complex reasoning, document analysis and agentic workflows.
The model arrived on August 13, 2026, only three weeks after Gemini 3.6 Flash. Google describes Gemini 3.7 Flash as its most intelligent workhorse model yet for coding and agents. The company says the release brings algorithmic improvements to the core reasoning system behind Gemini 3.6 Flash rather than introducing an entirely new base model.
The release gives developers a model with a 1 million-token context window, support for text, images, video, audio and PDF files, and a maximum output of 64,000 tokens. Gemini 3.7 Flash also supports function calling, code execution, search grounding, file search and computer use in preview.
Google is also launching the model with temporary API pricing of $0.75 per 1 million input tokens and $3.75 per 1 million output tokens. Those rates expire on December 31, 2026. From January 1, 2027, the prices rise to $1.50 and $7.50 respectively.
For developers, the release is less about having another chatbot and more about getting an AI model to handle larger, connected tasks.
What Is Gemini 3.7 Flash?
Gemini 3.7 Flash is the latest version of Google’s Flash model line.
Google DeepMind says the model is based on Gemini 3.6 Flash and introduces algorithmic improvements to its core reasoning foundation. Developers also get customizable thinking levels, allowing them to select low, medium or high reasoning depending on the task.
The model is available through Google AI Studio and the Gemini API, along with several Google products and platforms. Google lists Gemini 3.7 Flash as generally available.
The model accepts text, images, video, audio and PDF files. This makes the system useful for tasks where information comes from several formats.
A developer working with a long PDF, for example, does not need to treat the document as plain text alone. Gemini 3.7 Flash is designed to process the document as part of a broader multimodal workflow.
The model also gives developers a large context window of up to 1,048,576 input tokens. The output limit is 65,536 tokens in Google’s API documentation.
Gemini 3.7 Flash Gets a Major Coding Upgrade
Coding is one of the biggest areas of improvement.
Google’s evaluation shows Gemini 3.7 Flash scoring 43.6% on FrontierCode 1.1 Main, compared with 34.4% for Gemini 3.6 Flash. The benchmark measures production code quality.
The model also scored 65.3% on DeepSWE v1.1, a benchmark focused on long-horizon software engineering. Gemini 3.6 Flash scored 48.6%.
On Terminal-Bench 2.1, which tests agentic terminal coding, Gemini 3.7 Flash scored 85.8%, compared with 78.0% for its predecessor.
Google also recorded a higher Code Arena score for web development. Gemini 3.7 Flash reached an Elo score of 1,588, up from 1,538 for Gemini 3.6 Flash.
These figures show where Google expects the model to compete.
Gemini 3.7 Flash is aimed at software developers who want AI to handle longer coding tasks, work through multiple steps and operate as part of a larger development workflow.
The model is also designed for web development. Google positions 3.7 Flash as a tool for building and improving web applications, with stronger instruction following and reasoning for complex development tasks.
Document Understanding Gets Better
Gemini 3.7 Flash also brings stronger document understanding.
Google’s model evaluation shows a 34.0% score on GDP.pdf, an expert PDF document comprehension benchmark. Gemini 3.6 Flash scored 22.0%.
The difference matters for users working with long reports, research documents and business files.
The model’s 1 million-token context window gives developers more room to provide large amounts of information in a single request. Gemini 3.7 Flash also accepts PDF files directly through the API.
This combination gives developers room to build systems for document search, summarization, information extraction and analysis.
For businesses, the same capabilities fit workflows involving large internal documents. An AI system could process a report, identify important sections and return the information in a structured format.
Gemini 3.7 Flash Pushes Google’s AI Agent Strategy
Google is also placing Gemini 3.7 Flash at the center of its agent strategy.
An AI agent needs to handle more than a single question. A system needs to reason through several steps, use tools and maintain context while working toward a goal.
Gemini 3.7 Flash supports function calling, code execution, file search and computer use in preview. Google also lists search and Google Maps grounding among its supported capabilities.
The model scored 30.4% on Google’s AutomationBench evaluation, compared with 17.0% for Gemini 3.6 Flash.
On Terminal-Bench 3.0, which measures broader agent capabilities, Gemini 3.7 Flash scored 14.9%, compared with 5.4% for Gemini 3.6 Flash.
These results explain Google’s focus on agentic workflows.
Instead of using AI only to generate text, developers are building systems where models interact with tools, files and software to complete connected tasks.
Gemini 3.7 Flash is designed for this type of work.
Independent Testing Shows Progress
Google’s own benchmarks are not the only evidence of improvement.
The Artificial Analysis Intelligence Index gives Gemini 3.7 Flash a score of 56, four points higher than Gemini 3.6 Flash at 52.
The score puts Gemini 3.7 Flash close to several competing models in the same evaluation. Google’s own comparison shows the model performing strongly in coding and agent tasks, while other models lead on some broader evaluations.
This is important because benchmark results do not mean Gemini 3.7 Flash is the best model for every task.
The stronger case for the model is its combination of coding performance, agent capabilities, multimodal input, large context and Flash-level pricing.
Google is targeting developers who need those features together.
Gemini 3.7 Flash Pricing
Google is giving developers an introductory price of $0.75 per 1 million input tokens and $3.75 per 1 million output tokens.
The introductory rates expire on December 31, 2026.
Starting January 1, 2027, the prices rise to $1.50 per 1 million input tokens and $7.50 per 1 million output tokens.
The pricing makes Gemini 3.7 Flash attractive for developers running applications with large numbers of model requests.
This is especially relevant for AI agents. Agent systems often make several model calls while completing a task. Token costs therefore become an important part of running the application.
Google is using the Flash name to position the model around this balance between intelligence, speed and cost.
The company is not presenting 3.7 Flash as a small model for simple questions. Google calls the model its most intelligent workhorse yet and highlights coding, agents, advanced reasoning and multimodal understanding as its main use cases.
What Nigerian Users Need to Know
The release is also relevant to Nigerian AI users, but access differs depending on the Google product.
Gemini 3.7 Flash is available through Google AI Studio and the Gemini API. Google also lists Gemini App, Gemini Enterprise App, Gemini Enterprise Agent Platform and Google Antigravity among its distribution channels.
Gemini Spark is a separate experience built around Google’s agent capabilities.
Recent reporting has highlighted regional limits around Spark, including restrictions affecting Nigeria. Users therefore should not assume access to every Gemini 3.7 Flash-powered feature simply because the underlying model has launched.
For Nigerian developers, Google AI Studio and the Gemini API remain the more relevant routes for building with the model, subject to Google’s regional and account requirements.
The distinction matters because model availability and product availability are not always the same.
Why Gemini 3.7 Flash Matters
Gemini 3.7 Flash shows how quickly AI model development is moving toward specialized work.
Google released Gemini 3.6 Flash in July and followed with Gemini 3.7 Flash three weeks later. The company has now put more emphasis on coding, agents, document understanding and other tasks where users expect AI to complete several steps.
The release also arrives as Google faces pressure from OpenAI and Anthropic.
Reuters reported that Google launched Gemini 3.7 Flash while its expected flagship Gemini 3.5 Pro still had no public release date.
That makes the Flash line more important to Google’s current AI strategy.
Gemini 3.7 Flash does not try to win every benchmark. Instead, Google is pushing a model designed to deliver strong reasoning while keeping speed and API costs relatively low.
For developers, the combination is significant.
A model with a 1 million-token context window, stronger coding scores, document understanding, tool use and agent capabilities gives developers more options for building practical AI applications.
Conclusion
Google Gemini 3.7 Flash is a major upgrade to Google’s fast model line, with its strongest gains appearing in coding, agentic workflows and document understanding.
The model scored 43.6% on FrontierCode 1.1 Main, 65.3% on DeepSWE v1.1 and 85.8% on Terminal-Bench 2.1. Google also recorded a 34.0% result on GDP.pdf for document comprehension, compared with 22.0% for Gemini 3.6 Flash.
Developers also get support for a 1 million-token context window, PDF and multimodal inputs, function calling, code execution, file search and computer use in preview.
The introductory API price of $0.75 per 1 million input tokens and $3.75 per 1 million output tokens gives the model another advantage through the end of 2026.
For AI developers and businesses, Gemini 3.7 Flash is therefore less about another chatbot release and more about building systems capable of handling larger, longer and more complex tasks.
Frequently Asked Questions
What is Gemini 3.7 Flash best for?
Google positions Gemini 3.7 Flash for coding, agentic workflows, advanced reasoning, multimodal understanding and knowledge work.
How much does Gemini 3.7 Flash cost?
The introductory price is $0.75 per 1 million input tokens and $3.75 per 1 million output tokens through December 31, 2026. The rates increase to $1.50 and $7.50 respectively from January 1, 2027.
Does Gemini 3.7 Flash support PDFs?
Yes. Gemini 3.7 Flash accepts PDF files alongside text, images, video and audio. Google gives the model a 1 million-token input limit and a 64K output limit.
