Gemini New Model: What’s New in 2026?
Google has introduced another major update to its artificial intelligence ecosystem with the Gemini new model lineup. The latest generation includes Gemini 3.8 Flash, Gemini 3.8 Flash Cyber, Gemini 3.8 Live, and Gemini 3.8 Live Extended Thinking.
The latest releases show that Google is moving Gemini beyond traditional chatbot experiences. The company is focusing heavily on reasoning, coding, autonomous AI agents, real-time voice conversations, cybersecurity, and multimodal interaction.
For developers and businesses, this means the new Gemini models are designed not only to generate answers but also to work through complicated tasks, interact with tools, and support longer AI workflows.
What Is the Gemini New Model?
The Gemini new model refers to Google’s latest generation of Gemini AI models, with Gemini 3.8 Flash being one of the most important recent releases.
Google introduced Gemini 3.8 Flash on September 2, 2026. The company describes it as its most intelligent workhorse model, with improvements in software engineering, agentic tasks, and complex multi-step reasoning compared with Gemini 3.7 Flash.
The Gemini 3.8 family has since expanded into real-time conversational AI. On September 15, Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking for voice-first applications and more sophisticated live dialogue.
Key Gemini 3.8 models include:
- Gemini 3.8 Flash – focused on reasoning, coding, and AI agents
- Gemini 3.8 Flash Cyber – designed for cybersecurity applications
- Gemini 3.8 Live – designed for real-time voice interaction
- Gemini 3.8 Live Extended Thinking – designed for complex live reasoning
- Gemini 3.8 Flash TTS – focused on expressive text-to-speech generation
- Gemini 3.8 Flash-Lite TTS – optimized for high-volume voice generation
Google’s expanding model family demonstrates how Gemini is becoming a broader AI platform rather than a single chatbot model.
Gemini 3.8 Flash Features
The biggest headline in the Gemini new model lineup is Gemini 3.8 Flash.
Google says Gemini 3.8 Flash delivers significant improvements over Gemini 3.7 Flash while maintaining the speed and relatively low cost associated with the Flash family. The model is designed for applications where developers need a combination of intelligence, responsiveness, and scalability.
Major Gemini 3.8 Flash features include:
- Advanced multi-step reasoning
- Software engineering capabilities
- Long-horizon coding
- Agentic task execution
- Iterative tool use
- Professional knowledge workflows
- Improved prompt-injection robustness
- Support for enterprise applications
- Specialized cybersecurity capabilities through the Cyber variant
Google reports that Gemini 3.8 Flash achieved 54.9% on HLE-Verified, a benchmark covering multidisciplinary reasoning across areas such as STEM, humanities, and professional knowledge.
Benchmark numbers should be interpreted carefully because real-world performance can vary according to the task, prompting strategy, tools, and application architecture.
Gemini 3.8 for Coding
Coding is one of the most important use cases for the new Gemini model.
Gemini 3.8 Flash is designed for software engineering workflows that can involve multiple steps rather than simply generating an isolated piece of code.
According to Google, the model performs strongly on long-horizon software engineering tasks, where an AI system must work through a complicated problem from beginning to end.
Developers can potentially use Gemini for:
- Writing application code
- Debugging existing programs
- Understanding large codebases
- Creating websites
- Building prototypes
- Automating development tasks
- Reviewing technical implementations
- Working with software development tools
- Creating AI-powered applications
This is an important shift in AI coding. Instead of treating an AI model as a simple code generator, developers can increasingly use it as an assistant that participates in a broader development workflow.
Gemini 3.8 and AI Agents
AI agents are another major focus of the latest Gemini AI model.
A traditional chatbot generally receives a question and produces an answer. An AI agent can take a more active approach by breaking a goal into smaller tasks, using tools, examining results, and continuing until the workflow reaches an intended outcome.
Gemini 3.8 Flash was specifically designed with these agentic workflows in mind.
Google says the model can perform additional reasoning steps and make iterative tool calls on complex tasks. This approach allows the model to spend more computational effort when a problem requires additional diligence.
Potential applications include:
- Research agents
- Coding agents
- Business automation
- Data analysis
- Customer-service systems
- Enterprise workflow automation
- Document processing
- Technical assistants
This agent-focused development could become one of the most important aspects of Google’s Gemini strategy.
Gemini 3.8 Live: Real-Time AI Conversations
The latest Gemini new model update is not limited to text and coding.
Google has introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking for real-time voice interactions. The models are designed to make conversations with AI more natural while allowing users to complete complex tasks through spoken interaction.
Gemini 3.8 Live emphasizes:
- Natural voice conversations
- Low-latency interaction
- Visual grounding
- Real-time task execution
- Conversational context
- Voice-agent development
The Extended Thinking version is aimed at more complicated requests where additional reasoning is useful.
For developers, these capabilities open the door to applications that behave more like real-time assistants rather than conventional text-based chatbots.
Gemini 3.8 Live With Live Avatar
Google has taken real-time Gemini interaction another step further with Gemini 3.8 Live with Live Avatar.
Announced on September 24, 2026, Live Avatar combines near-real-time video generation with Gemini’s live dialogue capabilities. Google describes the experience as a visual AI persona that can listen, see, and speak during a conversation.
This technology could be particularly useful for enterprise applications.
Possible use cases include:
- Virtual customer-service representatives
- Interactive training assistants
- Digital product guides
- Educational applications
- Enterprise support
- Interactive demonstrations
The addition of a visual avatar could make conversational AI feel more natural in situations where facial presence and real-time interaction are valuable.
Gemini 3.8 Flash Cyber
Google has also released Gemini 3.8 Flash Cyber, a specialized AI model designed for cybersecurity.
The company says the model focuses on areas such as vulnerability detection and automated patching. Access is being provided through Google’s Fairwind Program to trusted defenders, including government authorities, critical infrastructure operators, and software maintainers.
This specialized model demonstrates an increasingly common direction in AI development.
Rather than expecting a single general-purpose model to handle every professional requirement, companies are developing specialized variants for specific domains and deployment environments.
Gemini 3.8 Text-to-Speech
The Gemini ecosystem has also expanded into expressive audio generation.
Google recently introduced Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS. These models are designed to generate customizable voices and more expressive speech for applications such as podcasts, games, audiobooks, dubbing, and voice agents.
Google says the models provide controls for characteristics such as:
- Voice style
- Emotion
- Pacing
- Accent
- Character design
- Conversational timing
- Multi-speaker dialogue
This makes the latest Gemini platform increasingly multimodal, covering text, reasoning, voice, video, and interactive experiences.
Gemini 3.8 Pricing
Pricing is an important consideration for developers planning to build applications with the Gemini new model.
Google announced an introductory Gemini 3.8 Flash price of $0.75 per million input tokens and $3.75 per million output tokens. The company states that this introductory pricing ends on December 31, 2026. Starting January 1, 2027, the listed price becomes $1.50 per million input tokens and $7.50 per million output tokens.
Actual costs depend on factors such as:
- Input token volume
- Output token volume
- Model selection
- Tool usage
- Application architecture
- Caching
- Number of users
- Frequency of requests
For high-volume applications, developers should calculate expected token consumption before selecting a model.
Where Can You Use the New Gemini Model?
Google is making Gemini available across several products and developer platforms.
Gemini 3.8 Flash is available to developers through Google’s Gemini API and Google AI Studio, while Google also lists access through Android Studio and its enterprise offerings. Consumers can access 3.8 Flash through eligible Google AI Pro and Ultra subscriptions in products including the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets.
Developers can use Gemini for:
- AI applications
- Chatbots
- Coding assistants
- Research tools
- Voice agents
- Enterprise automation
- Content workflows
- Data analysis
The availability of different model variants gives developers more flexibility when choosing between speed, reasoning depth, voice interaction, and specialized capabilities.
Gemini New Model vs Previous Gemini Models
The evolution of Gemini shows a clear movement toward more capable and autonomous AI systems.
Earlier Gemini releases established Google’s multimodal AI capabilities. The Flash family then emphasized speed and cost efficiency, while later releases increasingly focused on coding, reasoning, and agentic workflows.
Gemini 3.8 Flash continues that progression by allowing the model to spend additional reasoning effort on difficult tasks. Google says it can use more tokens and make iterative tool calls when higher effort is required.
The latest Live models add another dimension by bringing these capabilities into real-time voice interaction.
In simple terms, Gemini is moving from answering questions toward helping complete tasks.
Why the Gemini New Model Matters
The significance of the latest Gemini release goes beyond another increase in benchmark scores.
The broader trend is the development of AI systems capable of working across longer workflows. Coding, research, business automation, voice interaction, and cybersecurity can all involve several connected actions.
The Gemini new model is designed around that reality.
Its combination of reasoning, tool use, multimodal capabilities, coding, voice interaction, and agentic workflows gives developers more options for creating applications that can perform useful work rather than simply generate text.
Frequently Asked Questions About the Gemini New Model
What is the latest Gemini model?
One of Google’s latest major Gemini releases is Gemini 3.8 Flash, introduced in September 2026. Google positions it as a workhorse model focused on reasoning, software engineering, and agentic workflows.
What is Gemini 3.8 Flash?
Gemini 3.8 Flash is a Gemini AI model designed for coding, complex reasoning, agentic tasks, and professional workflows. Google says it improves on Gemini 3.7 Flash while maintaining a focus on speed and cost efficiency.
Is Gemini 3.8 good for coding?
Google specifically designed Gemini 3.8 Flash for software engineering and long-horizon coding tasks. It can be used for code generation, debugging, application development, and broader software workflows.
What is Gemini 3.8 Live?
Gemini 3.8 Live is Google’s real-time conversational AI model. It is designed for natural voice interaction, visual grounding, and applications that require live dialogue.
What is Gemini Live Avatar?
Gemini 3.8 Live with Live Avatar adds a visual AI persona to real-time Gemini conversations. Google says it combines near-real-time video generation with speech and dialogue capabilities.
How much does Gemini 3.8 Flash cost?
Google’s introductory Gemini 3.8 Flash pricing is $0.75 per million input tokens and $3.75 per million output tokens. Google says those introductory rates expire December 31, 2026, with higher listed rates beginning January 1, 2027.
Can developers build AI agents with Gemini?
Yes. Agentic workflows are one of the central use cases for Gemini 3.8 Flash. Google specifically highlights tool use, iterative reasoning, and long-running workflows as capabilities of the model.
What is Gemini 3.8 Flash Cyber?
Gemini 3.8 Flash Cyber is a specialized Gemini model for cybersecurity. Google says it is designed for capabilities such as vulnerability detection and automated patching and is being made available to trusted defenders through its Fairwind Program.
Final Thoughts on the Gemini New Model
The Gemini new model represents a broader evolution of Google’s AI strategy. Gemini 3.8 Flash focuses on coding, reasoning, and autonomous workflows, while Gemini 3.8 Live expands Gemini into real-time voice interaction.
The arrival of Live Avatar and new Gemini 3.8 text-to-speech models pushes the platform even further into multimodal and interactive AI.
For developers, the biggest opportunity may be the combination of reasoning and tool use. Instead of using AI only to generate content, businesses can explore systems that analyze information, operate software, communicate with users, and complete multi-step workflows.
As Google’s Gemini family continues to expand, the competition in generative AI is increasingly moving toward AI agents, real-time interaction, multimodal intelligence, and practical automation. Gemini 3.8 is an important step in that direction.