Google's Gemini 3.7 Flash Is Better at Code and Costs Half as Much
Three weeks after Gemini 3.6 Flash, Google shipped 3.7 Flash with real coding gains and cut the launch price in half. The catch is in the calendar: that price expires on December 31.
Google released Gemini 3.7 Flash on Thursday, three weeks after its predecessor, and it arrives with two headlines: it is noticeably better at writing code, and it costs half of what Gemini 3.6 Flash cost at launch. Flash is Google’s workhorse line, the cheap and fast model meant for everyday jobs rather than the hardest ones. This is the most capable version so far.
The numbers Google published are mostly about coding and agents. On FrontierCode 1.1, a test of production-quality code, the model scores 43.6 percent against 34.4 percent for 3.6 Flash. On DeepSWE v1.1, which measures whether a model can carry a long software task through to the end without losing the thread, it climbs from 49.0 to 65.3 percent. Its rating on WebDev Arena, where humans vote on which model built the better web page, rose from 1538 to 1588. Google also reports gains in document comprehension and business process automation, and says its own measurements put 3.7 Flash ahead of both Claude Sonnet 5 and GPT-5.6 Terra. Those are Google’s benchmarks, so treat them as a starting point rather than a verdict.
Pricing is where it gets interesting. Launch rates are $0.75 per million input tokens and $3.75 per million output tokens, exactly half of what 3.6 Flash cost when it appeared. A token is the chunk of text a model reads and writes, roughly three quarters of a word. That price holds through December 31, 2026. On January 1 it doubles to $1.50 and $7.50. The model is available now through the API, AI Studio, and Antigravity.
What is actually going on here
Three weeks between releases is not normal product cadence, it is a price war fought with model numbers. The cheap tier is where the volume lives: chatbots, summarising, document processing, coding assistants running in the background. Whoever is cheapest per unit of usable work in that tier collects an enormous amount of usage, so Google is buying market share with an introductory rate and telling you upfront when it ends. Google credits “algorithmic improvements” for the jump in capability, which is plausible but unverifiable from the outside. What we can verify is the price sheet, and the price sheet says: cheap now, twice as expensive in January.
What this means for you: if you just use the Gemini app, nothing changes today. This is an API product, aimed at people and companies building things. If you are building something, the practical advice is to read the second date, not just the first. An app priced against the $0.75 rate is an app that becomes twice as expensive to run on January 1, 2027, which is close enough to matter for anything you are planning now. And if you use a coding assistant that lets you pick a model, 3.7 Flash is worth a try on real tasks. Benchmarks measure something, but a Thursday afternoon with your own codebase measures more.
Sources
Ling 3.0 Flash Is the Smartest Open Model of Its Size, and It Stopped Making Things Up
Ant Group's inclusionAI released Ling 3.0 Flash under the MIT license. The headline number is not the intelligence score, it is the hallucination rate falling from 97 to 44 percent.
