Tech

Google launches Gemini 3.7 Flash with pricing cut and coding upgrades

Google has released Gemini 3.7 Flash, an artificial intelligence model optimised for software engineering and agent workflows, introducing a 50 per cent reduction in introductory token costs alongside significant performance gains over its predecessor.

Editorial persona
Owen Mercer
Markets and Finance Editor
Published
Draft
Source: Hacker News · View original source
Tech
No image available
New AI model arrives just three weeks after predecessor, targeting enterprise developers and knowledge workers

Google has released Gemini 3.7 Flash, an artificial intelligence model designed for coding and agent workflows, featuring improved performance in software engineering, web development, and knowledge-dense fields such as finance and law. The release follows Gemini 3.6 Flash by three weeks, driven by developer feedback and algorithmic innovations. The model is available at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens. Gemini 3.7 Flash is being integrated into Gemini Spark for Google AI Pro and Ultra subscribers.

Performance benchmarks indicate that 3.7 Flash shows improved first-pass code accuracy and production-ready code generation compared to 3.6 Flash. Specific metrics show the FrontierCode 1.1 Main score increased to 43.6 per cent from 34.4 per cent, while the DeepSWE v1.1 score rose to 65.3 per cent from 49.0 per cent. In web development, the model achieved an Elo score of 1588 on Arena.ai’s WebDev Arena, up from 1538 for the previous version.

For knowledge work, the update delivers improved reasoning in finance, law, and biosciences. The GDP.pdf benchmark score rose to 34.0 per cent from 22.0 per cent, testing the model’s ability to process complex documents. Additionally, the AutomationBench score rose to 30.4 per cent from 17.0 per cent, demonstrating enhanced capability in completing real-world business workflows.

The model is being integrated into Gemini Spark for Google AI Pro and Ultra subscribers, improving efficiency for knowledge work and tool use within Google Workspace applications. This integration allows the personal AI agent to consolidate files, draft emails, and update status documents more effectively. Gemini Spark is available in over 160 countries.

Google’s Gemini app has surpassed one billion monthly active users, placing it on par with OpenAI’s ChatGPT. The release coincides with this milestone, marking the fastest-growing product in Google’s history. The model card for 3.7 Flash is available for further technical details.

Safety updates included in this release involve shipping with updated safeguards against misuse in Chemical, Biological, Radiological, and Nuclear (CBRN) domains and cyber offense. The introductory pricing is available through the end of the year.

Continue reading

More from Tech

Read next: Trump administration permits private firms to conduct international cyberattacks under federal oversight
Read next: OpenAI Chief Revenue Officer Denise Dresser to Depart Ahead of IPO
Read next: OpenAI Unveils ‘Ultrafast’ Mode for GPT 5.6 Sol in Partnership with Cerebras