Gemini 3.5 Flash-Lite
Gemini 3.5 Flash-Lite: Google's Most Cost-Efficient Gemini
Model Overview
Gemini 3.5 Flash-Lite is Google DeepMind's most cost-efficient model in the new July 2026 Gemini release batch, released on July 21, 2026 alongside Gemini 3.6 Flash and Gemini 3.5 Flash Cyber. Designed for high-volume, cost-sensitive deployments where economy and speed are the priority, it retains multimodal input capabilities and the 1M token context window from the broader Gemini 3.5 family while offering the lowest price per token of the batch.
✨ Key Features
| Feature | Description |
|---|---|
| Most Cost-Efficient | Lowest price per token in the July 2026 Gemini release batch |
| Multimodal Input | Accepts text, image, audio, and video inputs |
| 1M Token Context | Full long-context capabilities for document-heavy tasks |
| High Throughput | Optimized for high-volume, low-latency deployments at scale |
| Gemini Ecosystem | Full compatibility with Gemini API tools, function calling, and grounding |
🔗 Resources
| Resource | Link |
|---|---|
| Blog Announcement | blog.google |
| Try it in AI Studio | aistudio.google.com |
| Gemini API Docs | ai.google.dev |
📜 License & Access
Proprietary — Available via Google AI Studio and the Gemini API.
Key Features
Most cost-efficient model in the new Gemini batch — lowest price per token
Multimodal input support: text, image, audio, and video
1M token context window for long-form processing
Optimized for high-throughput, cost-sensitive deployments
Part of Google's July 2026 Gemini release alongside 3.6 Flash and 3.5 Flash Cyber
You might also want to compare
Verified Sources
Tags
Model Specs
Parameters
Undisclosed
Context Window
1M tokens
License
Proprietary
Deployment
Resources & Links
Lineage
Predecessor
Gemini 3.1 Flash-LiteCurator Notes
Released July 21, 2026 as the lightweight complement to Gemini 3.6 Flash. Exact pricing and benchmark scores TBD from Google's official pricing page.
Compare Specs
Compare parameters, context windows, modalities, and benchmark scores of this model side-by-side with others.
Compare Model