Gemini halves prices, Sheets builds apps: August 14 brief

Google has cut the introductory price of its newest fast AI model in half, a direct saving for businesses that pay to process large amounts of text. It also turned Google Sheets into a simple app builder. OpenAI is chasing near-instant answers, while Suno is giving businesses tighter control over AI-made audio.

Google halves the price of its newest fast model

Google released Gemini 3.7 Flash on August 13 at an introductory price of $0.75 per million input tokens and $3.75 per million output tokens through December 31. A token is a small piece of text that an AI reads or writes. Those rates are half the original prices for Gemini 3.6 Flash, which arrived only three weeks ago.

The official announcement also says Gemini Spark, Google's assistant that can carry out multi-step work, now uses the model for paid subscribers. Google reports better results on its selected tests for software work, document analysis, and business tasks. Treat those scores as vendor evidence until independent tests catch up.

This matters if your company pays for an AI connection inside customer support, document processing, or an internal assistant. The lower rate can reduce the bill, but only through year-end. Our marketing software audit shows how to find usage costs that get buried beneath subscription fees.

Google Sheets turns rows into working mini-apps

Google began rolling out Sheets canvas on August 13, letting people turn spreadsheet rows into interactive dashboards and boards with a plain-English request. Changes made in the visual canvas flow back to the original spreadsheet, and colleagues can share it like an ordinary sheet.

The Google product post says the English-language feature is available globally for Google AI Pro and Ultra subscribers. It is also rolling out to Workspace Business and Enterprise Standard and Plus accounts, along with the Google AI Pro for Education add-on.

For a small company, this can make an ugly sales tracker or job schedule usable without paying for a separate dashboard. Start with a copy of low-risk data, not payroll or sensitive customer records. Ask it to build one view that answers one weekly question, such as which jobs are late or which channel produced paid orders. Keep the original columns intact until the team trusts the new view. Our simple attribution setup explains the source fields a useful sales dashboard needs.

OpenAI makes its top model answer up to 14 times faster

OpenAI put GPT-5.6 Sol Ultrafast into limited preview on August 13, claiming speeds of up to 750 output tokens per second, or as much as 14 times its standard processing speed. The company says the service runs on Cerebras hardware and is available only to a select group while capacity grows.

Faster output matters most when a customer is waiting during a voice call, live support chat, or checkout. Most owners can ignore this preview today because OpenAI has published no Ultrafast price and broad access has not arrived. Speed also does nothing to correct a wrong answer. If you build customer-facing AI, measure the full time from the customer's question to a checked response, then compare the new tier when its price and access become public. The broader small-business chatbot comparison covers where a person should remain in the loop.

Suno gives AI music a proper editing desk

Suno launched Studio 2.0 on August 13 with MIDI editing, a chat-based production helper, track separation, audio effects, and automation. MIDI is a set of digital instructions for musical notes, while track separation pulls vocals and instruments apart so each can be edited. The release is limited to Premier subscribers.

The business use is control. A team making podcast beds, event music, or short campaign audio can now adjust individual parts instead of regenerating an entire song after one bad section. Suno says Premier users can export high-quality tracks without a limit, although it may later charge credits for creating custom effects. Before using any generated music publicly, keep the prompt, exported files, subscription record, and approval notes together. That paper trail supports the review process in our AI content policy template.

Worth watching

• Google's Gemini 3.7 Flash discount ends December 31, when its listed input and output rates are due to double.
• OpenAI has not announced Ultrafast pricing or a date for broad access, so the business case is still incomplete.
• Suno says custom effect creation uses no credits at launch, but a credit charge may appear later.