Google just rolled out Gemini 3.7 Flash, and Gemini Spark gets one of the biggest pieces of this update.
Starting today, Google is moving Spark over to the new model, which sharpens its tool use and makes it noticeably better at handling tasks that span multiple steps. Google specifically points to tasks like pulling together scattered files, drafting emails, and keeping status documents updated across Google Workspace.
Gemini 3.7 Flash: An Improvement to How Spark is Handling Tasks
Gemini Spark already keeps working after you close your laptop or lock your phone, and it operates across Gmail, Drive, Docs, Calendar, Keep, and Tasks.
Google recently extended that reach into Chrome too. Spark can use accounts you’re already signed into, along with saved passwords in your browser, to navigate websites, compare flight options, book apartment viewings, and start the booking process on your behalf. It still stops and hands control back to you before anything sensitive happens, like an actual payment.
You may also like: Gemini App for Mac Adds Dictation and Screen-Aware AI
Gemini 3.7 Flash is meant to make all of this more dependable. Google says the model now spends more effort on planning and choosing the right tools, adjusts better when it hits a snag, and should need fewer retries or manual step-ins from you.
The benchmarks back that up. Gemini 3.7 Flash scored 30.4% on AutomationBench, a jump from the 17% Gemini 3.6 Flash managed. It also outpaced GPT-5.6 Terra, which scored 23.6%, and Claude Sonnet 5, which came in at 10.7%.
Google Gemini 3.7 Flash Better Than Spark
Document-heavy tasks saw a good jump too. Gemini 3.7 Flash scored 34% on GDP.pdf, well ahead of Gemini 3.6 Flash at 22%, Claude Sonnet 5 at 28%, and GPT-5.6 Terra at 24.7%.
The pricing is what makes these numbers stand out even more. Under introductory pricing through December 31, Gemini 3.7 Flash costs $0.75 per million input tokens and $3.75 per million output tokens, a fraction of what Claude Sonnet 5 charges at $2 and $10, or GPT-5.6 Terra at $2 and $12.
You may also like: Claude Gets 15 New App Connectors for iPhone and Android
Coding benchmarks improved too. Gemini 3.7 Flash scored 43.6% on FrontierCode, narrowly beating Claude Sonnet 5 at 42.7% and GPT-5.6 Terra at 41.3%, and it also gained ground over Gemini 3.6 Flash on DeepSWE.
For Spark, this update looks like a real step forward, particularly if the gains in tool use and multi-step planning cut down on failed tasks and reduce how often you need to step in manually.
Power users are still waiting on something bigger, though. Google’s more ambitious Gemini 3.5 Pro model got a preview at I/O back in May, but it missed its expected June launch and still hasn’t shipped.
