Google, continuing its intense competition with other major tech companies, has introduced a new AI model named Gemini Flash 3.8; a model aimed at developers that focuses on speed, multi-step reasoning capability, and lower cost.
Gemini Flash 3.8 is the third generation of Google's Flash model family and was introduced only about 6 weeks after the previous generation. With this model, Google aims to narrow the gap between fast, low-cost models and more powerful models for complex tasks.
The new model is specifically designed to handle complex software, programming, and multi-step tasks, and while offering advanced capabilities, it continues to focus on speed and low cost.
Text window of over one million tokens
One of the important features of Gemini Flash 3.8 is its very large text window. This model can process more than one million tokens in input and provide up to 65,536 tokens of output to the user.
Such capacity can be important for developing complex interactive applications, analyzing large volumes of information, and executing extensive software projects.
Google also provides the option to choose among three levels of reasoning: low, medium, and high. Thus, developers can balance response speed, compute power, and cost depending on the complexity of the task.
Powerful performance at lower cost
Google has launched Gemini Flash 3.8 with an introductory price of $0.75 per one million input tokens and $3.75 per one million output tokens. This promotional pricing is set to continue through the end of December 2026.
One notable point about this model is its ability to perform additional reasoning steps to increase response accuracy. Gemini Flash 3.8 can also use multiple software tools to perform complex tasks without directly charging the user for these extra steps.
Results from some technical tests also show the notable performance of this model. Gemini Flash 3.8 scored 73.7 percent on the DeepSWE v1.1 complex programming benchmark; a performance that places it within the range of some powerful paid models.
The model also scored 89.4 percent on the Terminal-Bench 2.1 test and 61.4 percent on the financial Vals Finance test.
Special edition for cybersecurity
Simultaneously with the introduction of Gemini Flash 3.8, Google also introduced another specialized model called Gemini 3.8 Flash Cyber.
This model has been developed exclusively for the field of cybersecurity and is made available to digital defense teams and governmental organizations through a program called Fairwind.
The main focus of Gemini 3.8 Flash Cyber is the automatic identification and remediation of software vulnerabilities. Such a capability can help security teams identify issues more quickly and reduce the time required to respond to cyber threats.
The release of this version indicates that Google, in addition to developing general-purpose models, is seeking specialized uses of AI in areas such as software security and cyber defense.
Lower-cost model, more opportunities for developers
One of the most important messages of the Gemini Flash 3.8 launch is Google's effort to reduce the cost of using advanced AI models.
The processing cost of powerful models is one of the main barriers to developing AI-based tools; especially for startups and small developers with limited financial resources.
A model like Gemini Flash 3.8 can allow developers to build more complex AI agents and automated applications without processing costs rapidly escalating.
This matter is particularly important in developing the next generation of AI Agents, because these tools require continuous processing and the use of various tools to perform multi-step tasks.
Gemini Flash 3.8 pricing will increase in the future
Of course, the current price of Gemini Flash 3.8 is not permanent. According to the announced plan, Google intends to raise the standard price of this model starting in 2027.
The input price will rise from $0.75 to $1.50 and the output price from $3.75 to $7.50 per one million tokens.
Despite this increase, the new model will still belong to the category of models that attempt to balance advanced capabilities, response speed, and usage cost.
AI competition enters a new phase
The release of Gemini Flash 3.8 shows that competition among major AI companies is no longer solely about building the most powerful possible model. Now speed, processing cost, and the ability to perform complex tasks have become equally important.
With the Gemini Flash family, Google is trying to offer a model suitable for widespread use in everyday products and applications; a model that on one hand can perform complex tasks and on the other keeps the cost of use manageable for developers.
If this trend continues, the main competition in the AI market in coming years will likely be over offering models that can deliver more intelligence at lower cost and higher speed to users and developers.