Home›Crypto›Market Structure›Google launches Gemini 3.7 Flash as an AI coding and a…

Google launches Gemini 3.7 Flash as an AI coding and agents model

Gemini 3.7 Flash is generally available and designed to run tasks with less human help, while OpenAI is offering its faster GPT-5.6 Sol Ultrafast tier only via a limited preview and waitlist.

Reported · Written up by our desk · Market Structure

Google has launched Gemini 3.7 Flash, a low-cost coding-and-agents AI model that is now generally available. The company positioned the release as a faster way to build software and run autonomous, multi step workflows with less human assistance.

In its rollout, Google said Gemini 3.7 Flash can accept up to a million input tokens, roughly 750,000 words, and return 64,000 output tokens. The model can process multiple media types, including text, images, video, audio, and PDFs, and it can call tools and control a computer.

Google also highlighted speed and efficiency gains in testing, saying it completed a test coding task in 2 minutes and 13 seconds, compared with more than 5 minutes on its prior Flash model. It further claimed the model is both more capable and more efficient, and said it is not sacrificing quality for speed.

Separately, Decrypt reports that OpenAI previewed GPT-5.6 Sol Ultrafast, described as a Cerebras-powered service tier that runs its most capable model up to 14 times faster. OpenAI said the tier initially launched in the API for a select group of customers, with expanded access intended as capacity grows.

More like this

Sources