Summary: Google shipped Gemini 3.7 Flash on August 13, 2026, three weeks after 3.6 Flash, and calls it its most intelligent workhorse model for coding and agents. Vendor-reported benchmarks include FrontierCode rising from 34.4% to 43.6% and DeepSWE from 49.0% to 65.3%. Introductory pricing is $0.75 and $3.75 per million input and output tokens through 2026, doubling on January 1, 2027.
Google shipped Gemini 3.7 Flash on August 13, 2026, a mere three weeks after 3.6 Flash. Tulsee Doshi, Senior Director of Product Management, posted the announcement for the Gemini team and kept the framing plain: “our most intelligent workhorse model yet for coding and agents.” A release every three weeks, paired with an entry price halved at launch, says plenty about how hard Google is pushing into the developer market.
Benchmarks: A Near Ten-Point Jump on Coding
Measured against 3.6 Flash, every figure Google published headed upward. FrontierCode 1.1 Main moved from 34.4% to 43.6%. DeepSWE v1.1 went from 49.0% to 65.3%. WebDev Arena Elo climbed from 1538 to 1588. GDP.pdf, which tests complex document processing, advanced from 22.0% to 34.0%. AutomationBench, a gauge of business workflow automation, rose from 17.0% to 30.4%.
The two largest improvements land exactly where agents do their real work: reading documents that resist easy parsing and carrying multi-step business processes through to the end. Those are also the tasks that let a workhorse model earn its keep, because they determine whether an agent can be trusted to run unattended. Google’s own account of the change is that the model “thinks more diligently” — steadier multi-step planning and tool use, tuned for software engineering, knowledge work, and web development. One caveat worth holding onto: these are vendor-reported numbers, so read the deltas as a signal rather than a guarantee.
Pricing: Halve It to Win Share, Double It in 2027
The pricing deserves the closest look. From now until December 31, 2026, the introductory rate stands at $0.75 per million input tokens and $3.75 per million output tokens — half of what 3.6 Flash originally cost per million tokens. Come January 1, 2027, those numbers become $1.50 and $7.50 respectively.
What Google is running is a textbook land-grab: pull developers’ workloads onto its platform at half price, then restore the list price six months later. Moving a workload takes engineering effort, and so does migrating away once agent pipelines, prompts, and evaluations have been shaped around one API. Whether to commit long-running workflows to a quote that doubles inside six months is now a calculation every developer has to run — at today’s scale, token bills are a genuine line item, and a doubling is no rounding error.
The arithmetic also bites harder for agents than for chat. An agentic run re-reads context, calls tools, and iterates, so tokens per task run high; a model sitting at $0.75 input and $3.75 output keeps long agent loops affordable in a way the list price may not. The introductory window amounts to a subsidy for discovering whether the economics hold up at your scale — which is precisely the moment to measure them.
Coverage: Developers, Enterprises, and Consumers at Once
Google put 3.7 Flash across its entire product line in a single stroke. Developers find it in Google Antigravity, the Gemini API through Google AI Studio, and Android Studio. Enterprises receive it in the Gemini Enterprise Agent Platform and the Gemini Enterprise app. For consumers, it becomes the model underneath Gemini Spark, Google’s always-on personal agent, serving AI Pro and Ultra subscribers in more than 160 countries — with a support-page exclusion list naming the EEA, the UK, Switzerland, and Nigeria. The consumer angle is subtler than it appears: an always-on agent spends tokens in the background around the clock, which puts the introductory price behind a workload that never truly stops.
Safety drew an update as well, adding safeguards against misuse in CBRN and cyber-offense domains, aligned with Google’s bioresilience approach and cyber program. Enterprise buyers care about this, since they need to know the frontier of the lineup is not being pushed down-market without guardrails.
What the Fast Cadence Signals
A Flash release every three weeks tells you Google’s center of gravity sits with the “fast, cheap, smart enough” workhorse, not an annual blockbuster. Developers gain because capability and price improvements arrive quickly, and the distance between the cheap tier and the flagship keeps shrinking. The cost is evaluation overhead: benchmarks you ran last month are stale this month, and swapping in a new model can quietly shift how an agent behaves.
The practical takeaway for 2026: before binding a long-term project to one vendor, treat the release cadence and the pricing deadline as contract terms. Pin the version, re-run your evals on every upgrade, and put a calendar reminder on January 1, 2027 — that is when the bill doubles.