Google launches Gemini 3.7 Flash AI model

Gemini 3.7 Flash artificial intelligence AI agents coding benchmarks Google Search

Google has released Gemini 3.7 Flash, a new “workhorse model” designed specifically for coding, autonomous AI agents, and complex business workflows. Released just three weeks after Gemini 3.6 Flash, the update focuses on significant performance gains in software engineering and reasoning. Benchmarks show notable improvements, including a jump to 65.3% on DeepSWE v1.1 and 43.6% on FrontierCode 1.1.

To improve developer accessibility, Google has introduced a significant price reduction, setting introductory API costs at $0.75 per million input tokens and $3.75 per million output tokens through the end of 2026. The model is being integrated into various platforms, including Google Search’s AI Mode for eligible subscribers, Android Studio, and GitHub Copilot.

Additionally, Google has introduced a feature allowing users to hide visible watermarks on AI-generated media created via Nano Banana, Omni, and Lyria. While visible watermarks can be toggled off in settings, all generated files will still contain invisible SynthID watermarks and C2PA provenance metadata for traceability. The rollout of these features is currently underway across multiple platforms.

Entities

DeepMind · Gemini · Gemini 3.7 Flash · GitHub · Google · Google DeepMind · OpenAI

Claims

What the coverage asserts, and how many sources carry each claim.

[○ 1 SOURCE] The introductory price for Gemini 3.7 Flash is $0.75 per million input tokens and $3.75 per million output tokens through the end of 2026.
[○ 1 SOURCE] Users can now toggle off visible corner watermarks for media generated by Nano Banana, Omni, and Lyria. techviral.net
[● 2 SOURCES] Google released the Gemini 3.7 Flash model on August 13, 2026. www.michael-bickel.de
[○ 1 SOURCE] The option to disable visible watermarks will not be available in countries where local laws mandate visible AI labeling. techviral.net
[● 2 SOURCES] The price of Gemini 3.7 Flash is 50% lower than the launch price of Gemini 3.6 Flash. www.michael-bickel.de
[○ 1 SOURCE] Gemini 3.7 Flash is a refinement of the 3.6 architecture rather than a new base model.
[○ 1 SOURCE] AI-generated media will still contain invisible SynthID watermarks and C2PA provenance metadata even if visible watermarks are disabled. techviral.net
[● 2 SOURCES] Gemini 3.7 Flash scores 65.3% on the DeepSWE v1.1 benchmark. www.michael-bickel.de