Google releases Gemini 3.1 Flash Lite with 1M context at $0.25 per million input tokens

Google released Gemini 3.1 Flash Lite on May 7, 2026, a high-efficiency multimodal model priced at $0.25 per million input tokens and $1.50 per million output tokens. The model features a 1,048,576 token context window and is priced at half the cost of Gemini 3 Flash, according to Google.

Core specifications

• Context window: 1,048,576 tokens
• Input pricing: $0.25 per million tokens
• Output pricing: $1.50 per million tokens
• Modalities: Text, image, video, audio, and PDF inputs
• Release date: May 7, 2026

Technical capabilities

Gemini 3.1 Flash Lite supports four distinct thinking levels: minimal, low, medium, and high. These levels allow developers to fine-tune the trade-off between API cost and model performance based on task complexity. Google designed the model specifically for low-latency, high-volume workloads.

The model is optimized for lightweight agentic workflows and simple data extraction tasks. According to Google, the model prioritizes responsiveness and API cost efficiency over maximum capability.

Thinking mode integration

The model supports reasoning-enabled requests through OpenRouter's reasoning parameter. Developers can access the model's step-by-step thinking process through the reasoning_details array in API responses. When continuing conversations, the complete reasoning_details must be preserved in message history for the model to maintain reasoning continuity.

Pricing positioning

At $0.25 per million input tokens, Gemini 3.1 Flash Lite is priced at 50% of Gemini 3 Flash's cost. The output token pricing of $1.50 per million represents a 6:1 output-to-input ratio, standard for Google's Flash tier models.

What this means

Gemini 3.1 Flash Lite fills a specific market gap for applications requiring multimodal understanding at scale where cost and latency are primary constraints. The four-level thinking system gives developers granular control over the reasoning-cost trade-off, unusual for a "lite" model tier. However, Google has not disclosed benchmark scores or parameter count, making it difficult to assess performance relative to competing models like GPT-4o mini or Claude 3.5 Haiku. The 1M context window matches Gemini 3 Flash, suggesting Google maintained context capability while reducing computational requirements elsewhere in the model architecture.