Unlocking the Potential: Google’s Gemini 3.6 Flash Reducing Enterprise Agent Token Costs

Google Releases Gemini 3.6 Flash and 3.5 Flash-Lite for Enterprise AI Agents

Google has unveiled Gemini 3.6 Flash and 3.5 Flash-Lite, two new models aimed at reducing latency and token costs for enterprise AI agents. These models are designed to enhance the efficiency of autonomous software agents operating within production environments.

The Importance of Token Efficiency

Running autonomous software agents involves a delicate balance between task competency and token generation. Every additional token generated during a task adds to the overall cost and slows down the workflow. Google’s new models address this challenge by offering improved throughput and reduced token usage.

Introducing Gemini 3.6 Flash

Google’s latest model, Gemini 3.6 Flash, boasts 17% fewer output tokens compared to its predecessor, 3.5 Flash. This reduction in token usage has been validated through tests such as the Datacurve DeepSWE benchmark, where token usage dropped by up to 65%. With pricing set at $1.50/1M input tokens and $7.50/1M output tokens, Gemini 3.6 Flash is ideal for continuous reasoning loops.

In performance tests, Gemini 3.6 Flash has shown significant improvements over the older model. For instance, on the MLE Bench, the success rate increased from 49.7% to 63.9%. The model is particularly effective in real-world knowledge work scenarios, as demonstrated by its performance on Google’s GDPval-AA v2 test.

Applications in Figma, Hebbia, and Harvey

Companies like Figma, Hebbia, and Harvey have already integrated Gemini 3.6 Flash into their workflows. Figma, in particular, has leveraged the model to accelerate design iterations without compromising output quality. Legal platform Harvey and research tool Hebbia use the model for multimodal document work, showcasing its versatility.

Gemini 3.5 Flash-Lite for High-Volume Work

For high-volume document processing and agentic search tasks, Google offers Gemini 3.5 Flash-Lite. This model is optimized for efficiency, with pricing set at $0.3/1M input tokens and $2.5/1M output tokens. With a focus on speed and volume, Gemini 3.5 Flash-Lite delivers impressive results on tests like the GDM-MRCR v2 and GDPval-AA v2.

Introducing Gemini 3.5 Flash Cyber

Gemini 3.5 Flash Cyber is a specialized model designed for validating and remediating code vulnerabilities. It addresses the growing challenge of automated vulnerability scanners outpacing security teams in patching vulnerabilities. Google restricts the distribution of this model to government entities and vetted partners to prevent misuse.

Integration and Accessibility

Engineering teams can access these new models through the Gemini API via Google AI Studio, Android Studio, or the Gemini Enterprise Agent Platform. Consumers can also access the models through the Gemini app, with Gemini 3.5 Flash-Lite being integrated into Google Search.

Conclusion

Google continues to innovate in the field of AI with the release of Gemini 3.6 Flash and 3.5 Flash-Lite, offering improved efficiency and reduced token costs for enterprise AI agents. These models cater to a range of use cases, from high-volume document processing to code vulnerability remediation, showcasing Google’s commitment to advancing AI technology.