Google launches Gemini 2.0 Flash-Lite: Ultra-low-cost AI model officially launches

Recently, Google launched its most economical model, Gemini2.0Flash-Lite, and it is officially available for production purposes. Gemini2.0Flash-Lite is part of the Google Gemini family and positioned as the most cost-effective option, the model offers public previews on Google AI Studio and Vertex AI, with the goal of serving developers who need cost-effective AI solutions.

Its design emphasizes lightweight and is suitable for teams and startups on a budget, especially when it comes to handling large-scale text output tasks.

Pricing is a highlight of this model. Gemini2.0Flash-Lite's input tokens are priced at $0.075 per million and output tokens at $0.30 per million. This pricing strategy makes it competitive in the market, for example, with lower costs compared to OpenAI's GPT-4o-mini (input $0.15/million, output $0.60/million).

In terms of performance, this model inherits the advantages of the Gemini series, with a context window of 1 million tokens, which can process a large amount of data. It outperforms Gemini1.5Flash in most benchmarks, maintaining the same speed and cost, especially for high-frequency tasks.

Gemini2.0Flash-Lite supports multimodal input, but compared to 2.0Flash, it does not support image or audio output, nor does it have advanced features such as "Search as a Tool" or "Code Execution as a Tool". This makes it more focused on text generation tasks, suitable for scenarios where fast, low-cost solutions are needed. It can generate single-line titles for about 40,000 photos, costing less than $1, showing how efficient it is in real-life applications.

Google's move is seen as a further expansion of its AI strategy, especially in the game with competitors such as OpenAI and Anthropic.