Gemini 3.1 Flash Lite Now Available on AI Gateway

Google's latest lightweight model, Gemini 3.1 Flash Lite, is now accessible through Vercel's AI Gateway. This addition expands the model options available to developers building with the Vercel ecosystem.

Performance Improvements

The model demonstrates notable improvements over its predecessor, Gemini 2.5 Flash Lite:

• Translation: Enhanced multilingual capabilities
• Data extraction: Better accuracy for structured data tasks
• Code completion: Improved programming assistance

Use Cases

Gemini 3.1 Flash Lite is best suited for:

• High-volume agentic tasks requiring fast inference
• Data extraction and transformation workflows
• Applications where latency and cost are primary constraints
• Budget-conscious deployments at scale

Implementation & Features

To integrate this model with the Vercel AI SDK, set the model identifier to google/gemini-3.1-flash-lite-preview. The model supports four thinking levels for extended reasoning:

• minimal: Fastest responses with minimal reasoning
• low: Light reasoning overhead
• medium: Balanced reasoning and speed
• high: Maximum reasoning capability

Developers can configure thinking behavior using the thinkingConfig option in provider settings, optionally including detailed thought processes in responses.

AI Gateway Benefits

Using Gemini 3.1 Flash Lite through AI Gateway provides:

• Unified API: Single interface across multiple models and providers
• Cost tracking: Detailed usage and cost monitoring
• Reliability: Built-in retries, failover, and automatic provider routing
• Observability: Integrated logging and performance analytics
• Bring Your Own Key: Support for custom API credentials