Google launches Gemini 3 Flash and expands direct conflict with ChatGPT

THE Google announced this Tuesday 17 the launch of Gemini 3 Flashnew version of sound artificial intelligence (AI) aimed at rapid responses, reduced costs and large-scale use. The company also confirmed that the model will become the default in the Gemini app and in the search engine’s AI mode, replacing Gemini 2.5 Flash for everyday tasks.

This strategy marks a new stage in the escalation of the conflict between Google and OpenAI for control of the generative AI market. By making Gemini 3 Flash the standard model for billions of users in its ecosystem, Google is banking on massive distribution to gain ground against the competition. ChatGPTwhich remains the main reference in the segment.

According to the company, Gemini 3 Flash combines the advanced reasoning capabilities of Gemini 3 Pro with faster response times and lower operating costs. In practice, this means a model that is less “heavy” than state-of-the-art models, but robust enough to handle complex questions, image, video and audio analyses.

In internal tests and widely used industry benchmarks, the Gemini 3 Flash performed close to that of more advanced models. In general reasoning assessments, the system easily outperformed its predecessor and came close to the results achieved by models such as the Gemini 3 Pro and the GPT-5.2, recently released by OpenAI.

Another strong point was performance on multimodal comprehension tests, which assess the ability to interpret text, image, video and sound in an integrated way. This type of skill is increasingly central in practical applications, such as analyzing short videos, reading visual documents, or interpreting user-submitted audio files.

For the general public, the change will be seen above all in the speed and variety of responses. Google says the new model better understands the intent of questions and tends to provide more visual answers, using images, tables and practical examples, especially in search.

In addition to text questions, Gemini 3 Flash allows users to submit videos, photos, audio recordings, or even hand-drawn sketches for suggestions, explanations, or analysis. The company also tested using the model to quickly create app prototypes using simple commands.

Close dispute with OpenAI

The launch comes at a time of high tension in the AI ​​race. In recent months, Google has ramped up the frequency of Gemini updates, with the launch of Gemini 3 Pro, new image and video generation models, and greater integration of AI into search, messaging, and productivity services.

On the other side, OpenAI, which has reached declare a “red alert” internallyresponded with GPT-5.2 and with improvements to the ChatGPT image generation systemin addition to reinforcing its message of growth in business uses.

Google, in turn, is banking on the reach of its ecosystem as a competitive differentiator. By integrating Gemini natively into the search engine and widely used applications, the company is quickly expanding the model’s visibility, including to people who are not actively searching for AI tools.

This movement also targets the business market. Companies like Figma, JetBrains, Salesforce and Workday are already using Gemini 3 Flash in their internal products and workflows. For developers, the model is now available through APIs and platforms like Vertex AI, with pricing tailored to repetitive, high-volume tasks.

Google executives say Flash’s goal is to function as a “working model,” capable of maintaining continuous operations without the high cost of more advanced systems. The company also claims that in certain tasks the new model consumes fewer resources than previous versions, which can reduce expenses with prolonged use.

Although it doesn’t directly mention its competitors, Google acknowledges that the conflict has forced the entire industry to evolve more quickly. Evaluation is now no longer limited only to the brute force of models, but also to efficiency, cost and the ability to scale without loss of quality.