Google ships Gemini 3.7 Flash while its flagship 3.5 Pro stays delayed

Google (NASDAQ: GOOG) has launched Gemini 3.7 Flash on Thursday, August 13. This is its second low-cost coding model in a matter of three weeks, despite delaying the Gemini 3.5 Pro it promised developers as far back as May.

For individuals and enterprise users banking on Google’s AI stack, they might be a bit disappointed, as Google’s recent focus has been on upgrading cheaper models rather than the flagship model many thought would compete with Anthropic & OpenAI. 

The company‘s pitch to users is that 3.7 Flash is the strongest workhorse model for coding and agents. It comes just 3 weeks after 3.6 Flash was launched, and the company says the new model was born out of internal optimization work and feedback from developers. 

Google has not stated whether newer versions of Gemini 3.5 Pro would be released. 

Google built Gemini 3.7 Flash as a low-cost coding model 

Google’s main selling point is price. The new model will cost $0.75 per million input tokens and $3.75 per million output tokens from now till then end of the year. Gemini 3.7 Flash will cost half of what Google 3.6 Flash cost when it was released.

The price cut aims to compete in a market where competitors are dropping their prices. OpenAI currently prices the Luna version of its GPT-5.6 lineup at $0.20 per million input tokens and $1.20 per million output tokens, cheaper than Google’s 3.7 Flash.

Benchmarks move, but only just 

Google backed the launch with a set of benchmark gains over 3.6 Flash. On coding, the DeepSWE v1.1 test increased from 49.0% to 65.3%, and FrontierCode 1.1 Main rose from 34.4% to 43.6%, according to Senior Director Tulsee Doshi. The model’s WebDev Arena score moved from 1,538 to 1,588.

Its knowledge work went up a notch as well. The GDP.pdf benchmark measures how a model manages complex documents, and based on the benchmark, 3.7 Flash scored 34.0%, a lot better than the older model’s score of 22.0%. 

AutomationBench tests a model’s ability to handle everyday business workflows, and 3.6 Flash moved from 17.0% to 30.4%.

Who can use the new model?

Not everyone has access to the new model. Developers can use 3.7 Flash via the Gemini API, AI Studio, Android Studio, Google Antigravity, and the Gemini Enterprise platform. As for individuals, not everyone will have access; only Google AI Pro and Ultra subscribers will have access.

3.6 Flash still has the standard Gemini chatbot, and the upgrade to 3.7 has made it easier for Spark to handle multi-step tasks across Gmail, Google Docs, and other Workspace apps.

Where Google ranks in the AI race 

The larger context surrounding the launch is not encouraging. Gemini 3.7 Flash trails behind other major AI models on the Artificial Analysis Intelligence Index, with Anthropic’s Claude Opus 5 leading the way.

If you're reading this, you’re already ahead. Stay there with our newsletter.