Google Ships Gemini 3.7 Flash Just Three Weeks After Its Last Model, Still Silent on Its Delayed Flagship
Google released Gemini 3.7 Flash Thursday, a cheaper coding-focused model, while its flagship Gemini 3.5 Pro remains unshipped months after its promised June launch.
Google released Gemini 3.7 Flash Thursday, a cheaper coding-focused model, while its flagship Gemini 3.5 Pro remains unshipped months after its promised June launch.
Google released Gemini 3.7 Flash on Thursday, a coding- and agent-focused AI model that arrived just three weeks after Gemini 3.6 Flash, a model which CEO Sundar Pichai highlighted in quarterly earnings.
Google said the rapid release was driven by developer feedback and algorithmic improvements that will carry into future models.
The launch comes at half the introductory price of its predecessor and lands as Alphabet’s DeepMind division works to close a widening capability gap with Anthropic and OpenAI.
This gap is more noticeable as Google remains silent on the delay of its flagship model, Gemini 3.5 Pro, promised for June and still unreleased
A Workhorse Model Racing Ahead of Its Own Flagship
Google’s blog post called 3.7 Flash its most intelligent workhorse model yet for coding and agents, highlighting improvements in software engineering, web development, and knowledge-heavy fields such as finance, legal tasks, and biosciences.
On the GDP.pdf document comprehension benchmark, it nearly doubled its predecessor’s score, rising from 22.0% to 34.0%.
Reuters reported that Google is pitching the model specifically as a lower-cost option for businesses building autonomous AI systems capable of planning tasks and completing multi-step workflows with less human oversight.
This positions 3.7 Flash squarely against the kind of agentic coding tools Anthropic’s Claude Code and OpenAI’s Codex have already built strong reputations around.
The Flagship That Keeps Missing Its Own Deadlines
Reuters also connected Thursday’s release to a story that’s becoming harder for Google to avoid: Gemini 3.5 Pro, the premium model investors have been watching as a test of whether DeepMind can keep pace with its rivals, remains unshipped.
Google said in July that the model was being tested with partners and would arrive “soon.” That delay isn’t happening in isolation.
Reuters noted Google announced a sweeping leadership overhaul of DeepMind just last week, with chief Demis Hassabis, who recently became vocal about US-led AI governance, stepping aside for deputy Koray Kavukcuoglu.
At the same time, the two original technical co-leads of Gemini left the company entirely to start their own venture, a level of turnover that rarely accompanies a product on schedule.
Why Shipping Fast on Flash Doesn’t Fix the Real Problem
What’s important here is the strategic logic behind Google leaning into Flash while Pro stalls.
9to5Google reported that 3.7 Flash isn’t just cheaper; it follows instructions more reliably and requires fewer retries in engineering workflows, making it more useful in production than its benchmark results alone suggest.
But that also reveals the gap: Google is improving the model tier it can ship reliably, while the model meant to show DeepMind can match Anthropic’s Opus or OpenAI’s Sol tier keeps slipping.
Even the pricing complicates Google’s positioning: OpenAI’s comparable GPT-5.6 Luna costs $0.20 per million input tokens versus $0.75 for Gemini 3.7 Flash.
Google’s “lower cost” pitch is therefore about capability per dollar, not the lowest price.
A three-week release cadence shows engineering momentum, but it is concentrated in the tier that does not compete directly at the frontier where Google’s rivals have been pulling ahead.
Source: Introducing Gemini 3.7 Flash
