Google Enhances Gemini 2.5 LLM Series with New Entry-Level Model and Updated Pricing

Google LLC has launched Gemini 2.5 Flash-Lite, a new large language model (LLM) that enhances prompt processing speed and cost-efficiency over its predecessor. This is part of an update to the Gemini 2.5 LLM series, which now includes the general availability of both Gemini 2.5 Flash and Gemini 2.5 Pro. The models utilize a mixture-of-experts architecture that reduces hardware usage by activating only one neural network per prompt. Gemini 2.5, trained on Google’s TPUv5p chip, supports up to one million tokens and demonstrated superior performance in internal tests against OpenAI’s models. Flash-Lite is positioned as the new entry-level model, optimized for tasks like translation and classification, with significantly lower costs—$0.1 per million input tokens compared to $10 for Gemini 2.5 Pro. Additionally, pricing for Gemini 2.5 Flash has been adjusted as part of this update, accommodating broader accessibility for developers.

Source link