Google launches Gemini 3.7 Flash for coding, AI agent projects
Google LLC today launched its most capable entry-level artificial intelligence model yet. Gemini 3.7 Flash is rolling out three weeks after its predecessor. Despite the short release cycle, Google engineers managed to implement significant output quality improvements. The company says that Gemini 3.7 Flash outperformed comparable models from Anthropic PBC and OpenAI Group PBC across […] The post…
Google has unveiled Gemini 3.7 Flash, its latest entry-level AI model, just three weeks after its predecessor. Despite the rapid release, Google engineers managed to enhance the model's output quality significantly. According to Google, Gemini 3.7 Flash outperformed comparable models from Anthropic and OpenAI across nine benchmarks.
The AI excelled in FrontierCode 1.1 Main, a benchmark featuring 100 programming tasks in various languages, requiring not only code generation but also bug testing and adherence to project-specific style guides. Compared to the previous version, Gemini 3.7 Flash is particularly proficient at creating user interfaces, producing layouts that more closely match user-uploaded reference images.
The model can process up to 1 million tokens of images, video, and text per prompt, with responses up to 64,000 tokens of text. Google's senior director of product management, Tulsee Doshi, praised the model for its improved developer experience, emphasizing its better adaptation to challenges, clearer intent clarification, and greater adherence to instructions.
In a GDP.pdf benchmark, where AI models answer questions about business documents, Gemini 3.7 Flash achieved 34% accuracy, surpassing Claude Sonnet 5 (30%) and GPT-5.6 Terra (23%) by 6% and 9.3%, respectively. Moreover, Google highlights that Gemini 3.7 Flash can support AI agent ensembles, although the model's architecture remains undisclosed.
The model card indicates that it is based on the same architecture as Gemini 3.6 Flash, which is derived from Gemini 3 Pro, featuring a transformer-based mixture-of-experts architecture. However, details about the training process of Gemini 3.7 Flash are not available. Developers can access Gemini 3.7 Flash at half the price of Gemini 3.6 Flash until the end of the year.
Additionally, Google is incorporating the model into Gemini Spark, a consumer-focused AI agent that can browse the web and perform actions within other Google services.
Written by urgent.news from SiliconANGLE's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
