Key Takeaways
- Google announced Gemini 3.7 Flash on August 13 (US time), only three weeks after Gemini 3.6 Flash
- Google calls it its most intelligent "workhorse" model yet for coding and AI agents, with large jumps on software-engineering benchmarks
- Introductory API pricing is half of 3.6 Flash's launch rate: $0.75 (around ₱46) per million input tokens and $3.75 (around ₱230) per million output tokens, until December 31, 2026
- It is rolling out in the Gemini app through Spark for AI Pro and Ultra subscribers, plus AI Studio, Android Studio, Google Antigravity, and Gemini Enterprise
Google is not slowing down its release pace. Gemini 3.7 Flash arrived barely three weeks after 3.6 Flash, and 9to5Google reports the new model immediately replaces its predecessor inside Spark, the round-the-clock personal AI agent Google launched in the Gemini app earlier this year.
What is actually better
Google is pitching 3.7 Flash at programmers and businesses that automate work with AI agents — software that plans and carries out multi-step tasks on its own. The company says the model resolves bugs and completes coding tasks more reliably, builds more functional web layouts in fewer prompts, and reasons more accurately in fields like finance, law, and biosciences.
Benchmark gains at a glance
| Benchmark | Gemini 3.6 Flash | Gemini 3.7 Flash |
|---|
| DeepSWE v1.1 (software engineering) | 49.0% | 65.3% |
| FrontierCode 1.1 Main (coding) | 34.4% | 43.6% |
| WebDev Arena (Elo rating) | 1,538 | 1,588 |
| GDP.pdf (document understanding) | 22.0% | 34.0% |
| AutomationBench (agent tasks) | 17.0% | 30.4% |
The DeepSWE jump — from roughly half of tasks solved to about two-thirds — is the number Google is leading with, and it is the kind of gain that usually took a full model generation, not a three-week refresh.
Pricing: half now, double later
Through the API, Gemini 3.7 Flash launches at an introductory $0.75 (around ₱46) per million input tokens and $3.75 (around ₱230) per million output tokens — half of what 3.6 Flash cost at its debut. The promotional rate runs until December 31, 2026; from January 1, 2027, the listed standard price rises to $1.50 (around ₱92) and $7.50 (around ₱460) respectively. A token is the small chunk of text AI models read and write — a million tokens is roughly 750,000 words — so these rates translate to centavos for typical app requests.
Aggressive introductory pricing has become the front line of the AI race: rivals have been undercutting each other for months, as when Alibaba priced Qwen3.8-Max below Claude Opus 5.
Where you can use it
In the consumer Gemini app, 3.7 Flash powers Spark, which requires a paid Google AI Pro or Ultra subscription; the agent pairs the new model with tool use for Google Workspace apps such as Gmail and Docs, so it can act on your accounts rather than just chat. Developers get it immediately through the Gemini API in Google AI Studio and Android Studio, in the Google Antigravity development environment, and through the Gemini Enterprise Agent Platform and Gemini Enterprise app for companies.
What it means for Filipino developers
There is no Philippines-specific rollout to wait for here: API access through AI Studio is available to local developers the same day as everywhere else, billed in US dollars at the rates above. The half-price window makes the next four and a half months the cheapest time to test agent-style features in local apps before rates double in January. That lands in a market already leaning in — a recent Microsoft study found Filipino AI users rank among the world's most advanced — though the consumer-side Spark agent remains behind a paid subscription.
