Google Cuts AI Costs with Gemini 3.5 Flash

Google Tackles Token Costs with Gemini 3.5 Flash

Google recently unveiled Gemini 3.5 Flash, a strategic advancement aimed at reducing costs associated with tokens in generative AI. This cost is often cited as one of the main obstacles to the widespread adoption of AI in businesses. In response, Google has drastically lowered its prices and increased the number of AI agents integrated into its Workspace and Google Cloud platforms. The goal is to make AI more economically viable, especially in the face of competitors like OpenAI, Anthropic, and open source models.

Companies are facing rising expenses due to the increasing number of AI agents in their professional tools. These expenses, primarily driven by tokens, are beginning to exceed certain annual budgets. With its Gemini 3.5 Flash model, Google aims to address this issue. This model was presented at the I/O 2026 conference.

An Attractive Business Model for Companies

Gemini 3.5 Flash is presented by Google as its most cost-effective offering to date, with a price set at $1.50 per million tokens. This pricing aims to convince businesses that AI can be profitable at scale. By reducing costs, Google hopes to facilitate the adoption of agent-based AI.

Responding to the Surge in AI Agent Costs

For several months, companies have been exploring the use of AI agents to automate complex tasks such as document research, email management, internal support, data analysis, and workflow orchestration. However, each interaction with these agents consumes tokens, and the volumes are increasing rapidly. Sundar Pichai, CEO of Google, emphasized that "some organizations have already consumed their annual token budget while the year is barely underway." This issue is becoming critical for IT departments and finance teams.

With Gemini 3.5 Flash, Google promises a faster, cheaper model that is sufficiently powerful to support large-scale daily use. This approach is reminiscent of Anthropic with its Claude Opus 4.6 model, which has also seen its prices drop.

An Integrated and Cost-Effective Agent-Based AI

The launch of Gemini 3.5 Flash is accompanied by several new products aimed at strengthening Google's agent ecosystem, including Gemini Spark and Omni Flash. Gemini Spark is presented as a personal agent that operates in the background, capable of directly accessing Gmail, Docs, Sheets, or Slides to perform various tasks for users. This agent can compile information, organize events, send reminders, and update spreadsheets in real time.

Google emphasizes the native integration with Google Workspace, which limits the need for external APIs, thereby reducing token-related costs. This strategy aims to optimize new always-on use cases, where agents operate continuously within employees' work environments. In this context, saving a few cents per million tokens can represent millions of dollars for a large corporation.

Google also introduced AntiGravity 2.0, a new version of its multi-agent platform. According to the company, this infrastructure enabled the creation of a complete operating system in just 12 hours for less than $1,000 in API credits. This demonstration aims to show that it is possible to develop complex AI applications without requiring enormous budgets.

Intense Economic Competition in the AI Sector

The announcement of Gemini 3.5 Flash marks a new phase of intense economic competition in the generative AI market. For nearly two years, the battle has focused on the power of models. Now, companies are demanding more visibility on costs, profitability, and operational stability.

Google seeks to stand out against OpenAI, Anthropic, and rapidly advancing open source models. Each new launch becomes a strategic response to a competitor.

However, this acceleration creates challenges for companies. Models are evolving quickly, forcing technical teams to continuously test new versions, review their architectures, and adapt their tools.

For many organizations, the real challenge is to standardize generative AI sustainably without letting costs spiral out of control. With Gemini 3.5 Flash, Google demonstrates that the future of enterprise AI relies on the ability of models to deliver a credible return on investment.