All newsReleases

Google's new Flash model is "half price" — and the price list says that ends on 31 December

Published: 8/15/2026 · Source: Google (blog and Gemini API price list)

Google announced Gemini 3.7 Flash on 13 August 2026 and called it "our most intelligent workhorse model yet for coding and agents". The benchmark table backs the coding half of that claim. Against Gemini 3.6 Flash, released three weeks earlier, FrontierCode 1.1 rises from 34.4% to 43.6%, DeepSWE v1.1 from 49.0% to 65.3%, AutomationBench from 17.0% to 30.4% and the document-reading test GDP.pdf from 22.0% to 34.0%. WebDev Arena puts it at 1588 Elo against 1538. Those are large single-generation jumps for a model line that Google has until now sold on latency rather than depth. The part worth reading twice is the price. Google presents USD 0.75 per million input tokens and USD 3.75 per million output as "an introductory price of half the original 3.6 Flash cost". The word doing the work is "introductory". In Google's own price list the rate carries an expiry date: it holds through 31 December 2026, and on 1 January 2027 input goes to USD 1.50 and output to USD 7.50. That is not a discount on a permanent price — it is a countdown to the permanent price. There is a second detail the announcement leaves out. Gemini 3.6 Flash carries the identical discounted rate today, with the identical expiry date. So as of 15 August 2026 the new model and the one it replaces cost exactly the same to run, and both double at the new year. Anyone budgeting a 2027 workload on today's figures will be out by a factor of two. The technical envelope is unchanged from the rest of the Flash line: a context window of 1,048,576 tokens in and 65,536 out, text, image, video, audio and PDF accepted, text returned. Thinking is offered at low, medium and high levels — the "minimal" setting available elsewhere in the family is not supported here. The model ships as stable in Google Antigravity, the Gemini API through AI Studio and Android Studio, the Gemini Enterprise Agent Platform, and Gemini Spark for AI Pro and Ultra subscribers in more than 160 countries. Editorial note: the benchmark figures and the "half price" wording come from Google's announcement; the expiry date and the 3.6 Flash comparison come from Google's published price list, which we read directly.