A quick look at Gemini API shinny new pricing plan.


Google AI Gemini API has been gradually gaining traction, and it’s now transitioning into mainstream adoption. Gone are the days when Google merely relied on developers to test its new AI platform. With the platform now deemed production-ready, Google is gearing up to enable billing, marking a significant shift in its approach for Gemini. EU developers can only hope that this move signals a broader availability of the Gemini API worldwide. Many countries have been excluded from testing the Google AI Gemini API, leaving developers in those regions without access to its capabilities. (see regions)

There has been ongoing speculation about Google’s timeline for implementing a payment tear for its powerful new AI tool, Google AI. While Vertex AI, with its payment tiers, is already established, developers have enjoyed free access to the Palm API, the initial API behind MakerSuite since its launch in September 2023. Now rebranded as Google AI Studio Gemini. Even with the remarkable release of the Gemini AI 1.5 pro model, boasting a 1 million token context window, usage and testing have remained free for developers as Google continued to improve this offering. This means that Google bears the sole cost of developer usage as we continue to refine and experiment with their new AI models, paving the way for the creation of applications leveraging AI without concern for backend expenses.

With the announcement of the two tear pricing plan for Gemini AI behind Google Ai Studio. Google is bidding farewell to the era of free access to its Gemini AI API. This move signals a strategic shift towards a more financially-driven model in its AI development endeavors.

Previously, developers enjoyed complimentary access to Google’s AI offerings as a means to attract them away from competitors like OpenAI. However, the landscape is evolving. OpenAI, having been the pioneer in this domain, has already monetized its APIs and granted access to large language models (LLMs). Now, Google aims to follow suit, intending to integrate similar monetization models into its cloud and AI Studio services. It appears that the days of unrestricted free access are drawing to a close.

Pricing Tiers

Free of Charge:

Google AI Studio’s Gemini API offers a remarkable benefit: it’s entirely free of charge. This means developers can access its powerful capabilities without incurring any financial costs. However, to maintain fair usage and system stability, there are rate limits imposed on API usage.

Rate Limits:

  • Requests per Minute (RPM): Developers are allowed up to 2 requests per minute to the API.
  • Tokens per Minute (TPM): A limit of 32,000 tokens per minute ensures that token-based operations remain within reasonable bounds. This will undoubtedly prevent developers from testing with the 1 million token context window of models like Gemini 1.5 pro
  • Requests per Day (RPD): Developers can make up to 50 requests per day, ensuring consistent access over time.

Prompts/Responses Used to Improve Products:

Participation in prompts/responses for product improvement. This is the key issue with usage in the EU currently. The EU does not want new AI models trained on EU data leaving the EU. I wonder wither this option will even be available in the EU once it is finally available in the EU.

Pay-as-You-Go:

Coming May 2, 2024 Google AI Studio’s Gemini API will offer a new pay-as-you-go model, offering flexibility and scalability to developers. While there are rate limits in place to ensure fair usage and system stability, users can access the API’s powerful capabilities based on their needs.

Rate Limits:

  • Requests per Minute (RPM): Developers are permitted up to 5 requests per minute to the API.
  • Tokens per Minute (TPM): The API allows for a maximum of 10 million tokens per minute, ensuring efficient token-based operations.
  • Requests per Day (RPD): Users can make up to 2,000 requests per day, providing consistent access over time.

Price (Input & Output):

For input, developers can expect to pay $7 per 1 million tokens based on preview pricing. Similarly, for output, the cost is $21 per 1 million tokens under the preview pricing model.

Prompts/Responses Used to Improve Products:

At this time, prompts/responses used to improve products. This model is similar to what Vertex AI already offers and will probably what the EU has been waiting for to grant Google AI Gemini access to the EU market.

Conclusion

If you are a developer out there and you expected to have access to Google AI Gemini for free forever then you were kidding yourself. This has always been a quick , easy and cheep way for Google to get US to test their new product.

I hope you have enjoyed it I know I have. Now lets get to work and develop some amazing applications.


About Linda Lawton

My name is Linda Lawton I have more than 20 years experience working as an application developer and a database expert. I have also been working with Google APIs since 2012 and I have been contributing to the Google .Net client library since 2013. In 2013 I became a a Google Developer Experts for Google Analytics.

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.