Google Unveils Gemini 3.7 Flash with 50% Price Cut for Coding and Agent Workflows

Google has launched Gemini 3.7 Flash, its latest AI workhorse aimed squarely at software developers and autonomous agents, and it arrives with a temporary ↑ 50% discount on API usage.

Gemini 3.7 Flash: coding boost and price slash

The upgrade follows a three‑week sprint after Gemini 3.6 Flash, a cadence Google attributes to developer feedback and algorithmic refinements. In benchmark suites the model jumps to 43.6% on FrontierCode, overtaking Claude Sonnet 5 and edging past GPT‑5.6 Terra, while its web‑development Elo climbs to 1588 on Code Arena. Enterprise workflow scores rise to 30.4% on AutomationBench, a leap from 17.0% a month earlier. Pricing runs at $0.75 per million input tokens and $3.75 per million output tokens through Dec 31 2026, then doubles to $1.50/$7.50 in 2027.

“The model thinks more diligently, applying extra effort to multi‑step planning and tool calls,” Google DeepMind said in a release.

The lower cost and higher first‑pass accuracy are meant to cut total cost of ownership for high‑volume coding bots and document‑processing agents. Google also embeds the model in the Gemini Spark personal assistant for AI Pro and Ultra subscribers, extending its reach into Google Workspace. Leadership shifts – with Demis Hassabis moving to Alphabet chief scientist and Koray Kavukcuoglu now steering Gemini development – signal a tighter product focus after the delayed Gemini 3.5 Pro, which Reuters reported missed internal targets. The move comes as the tech giant wrestles with internal talent churn and a broader debate over whether it can keep pace with rivals. For enterprises still recovering from the pandemic-induced digital acceleration, the price cut and performance gains may tip the balance toward Google’s ecosystem.

Reported by: Dr. Aris Thorne

Artificial Intelligence Researcher