Three weeks. That's how long Google gave Gemini 3.6 Flash to live this time. On August 13, Gemini 3.7 Flash went live, just about three weeks after its predecessor's release — and that pace alone is newsworthy. Most model makers are still working on quarterly release cycles; Google is now operating on weeks.
3.7 Flash sticks to a familiar playbook: multimodal support, a 1M-token context window, and up to 64K tokens of output, geared toward software engineering and agentic tasks. That positioning has become the battleground du jour across model makers, but this time Google's real weapon isn't capability — it's price.
The entry-level pricing sits at $0.75 per million input tokens and $3.75 for output, with a blended cost around $1.35 — and Google has already flagged that this pricing will rise starting in 2027. Stacked against its peers, that number undercuts both Sonnet 5 and GPT-5.6 Terra. In other words, Google's bet right now is: drive the price down first, then get developers hooked on building agentic tasks with Gemini.
Interestingly, OpenAI took a different route at the same time. They opened an invite-only preview of GPT-5.6 Sol Ultrafast, selling it on output speeds of 750 tokens per second. One company is fighting on price, the other on speed — two different answers to "what's the next thing to compete on," but both sidestepping the most expensive and hardest-to-deliver promise of all: being smarter.
All facts in this article are drawn from public release information. No official launch timeline or full pricing tiers have been provided; readers should watch for further official announcements.
