Google Releases Gemini 3.7 Flash With 50% API Price Discount Targeted at Coding and Agents

Google has introduced Gemini 3.7 Flash, an updated version of its primary lightweight artificial intelligence model, tailoring the release toward automated coding, agent-based workflows, and enterprise knowledge tasks while offering a temporary 50 percent price reduction on API access.

Under the promotional structure, developers using Gemini 3.7 Flash will pay $0.75 per million input tokens and $3.75 per million output tokens through December 31, 2026. On January 1, 2027, rates are scheduled to double to $1.50 per million input tokens and $7.50 per million output tokens. As VentureBeat first reported, the temporary pricing window gives enterprise software teams several months to test whether the model's performance improvements translate to reduced operational expenditures for high-volume automated workflows.

The arrival of version 3.7 comes just three weeks after Google deployed Gemini 3.6 Flash. Google credited the unusually tight turnaround to rapid algorithmic enhancements and direct feedback from the developer community. The accelerated update cycle highlights the search giant's focus on refining its mid-tier Flash product line while its next flagship tier remains in development.

Notably, Google has yet to attach a launch date to its anticipated Gemini 3.5 Pro model. Reuters previously noted that Google provided no release timeframe during Thursday's announcement, even though the Pro variant had earlier been characterized as undergoing testing with partner organizations. Axios similarly pointed out that the 3.7 Flash release precedes the rollout of the higher-tier Pro iteration.

Positioned by Google as its most capable mid-tier system for software development and autonomous agents, Gemini 3.7 Flash focuses on instruction adherence and task completion. The company reported that the system demonstrates enhanced adaptability when encountering logic errors, seeks clarification on ambiguous tasks more effectively, and limits unnecessary code alterations during automated programming tasks.

In enterprise deployment scenarios, such gains are designed to lower human intervention rates in automated systems. By executing complex multi-step instructions and recovering autonomously from execution faults, Google claims the upgraded system can minimize manual oversight and shorten the feedback loops required for enterprise coding and administrative agents to complete work successfully.