Recently, Google has made another significant move in the AI model price war, announcing a substantial price reduction of up to 78% for its fast Gemini1.5Flash AI model. This is undoubtedly good news for developers. According to Google's announcement, the cost for input tokens will be reduced to $0.075 per million tokens, and the cost for output tokens will be reduced to $0.30 per million tokens, applicable to prompts of up to 128,000 tokens. Even for longer prompts and caching, similar price adjustments will apply.
Gemini1.5Flash is widely used in scenarios requiring quick response and low latency, such as summarization, classification, and multi-modal understanding. Google's new API and AI Studio also support enhanced PDF understanding capabilities, enabling analysis of PDFs containing visual content such as graphics and images, showcasing its powerful multi-modal processing abilities.
In addition, Google has expanded language support for the Gemini1.5Pro and Flash models, now covering over 100 languages. This means developers from different regions can work in their familiar language environments, avoiding issues where responses are blocked due to unsupported languages. Google has also opened up the fine-tuning feature for Gemini1.5Flash to all developers. This feature allows developers to provide additional data for specific tasks, customizing the base model to enhance its performance. By doing so, developers can reduce the context size of prompts, thereby lowering latency and costs while improving model accuracy.
Google's announcement of this round of price cuts comes shortly after OpenAI announced a reduction of up to 50% in the access fees for its GPT-4o API. Clearly, the intense price competition in the industry is a response to the high costs of developing and operating AI models.
Key Points:
🤑 Significant Price Reduction: Google has reduced the input token price for Gemini1.5Flash to $0.075 and the output token price to $0.30, with the highest reduction reaching 78%.
🌍 Expanded Language Support: Gemini1.5Pro and Flash models now support over 100 languages, facilitating global development.
⚙️ Open Fine-Tuning: All developers can fine-tune the Gemini1.5Flash model via the API and AI Studio to enhance model performance and accuracy.