At I/O 2024, Google has announced some new AI models, upcoming projects and countless AI features will be on its products. However, it is worth noting that the Gemini 1.5 Flash model. This is a quick and impressive model, bringing multimodal ability and context time up to 1 million tokens (2 million through the waiting list).
Although Gemini 1.5 Flash has a small size – Google has not disclosed the parameters size – but it achieved high scores on all methods – text, images and sound. In Gemini 1.5 technical report, Google revealed that Gemini 1.5 Flash is much superior to larger models such as 1.0 Ultra and 1.0 Pro in many ways. Only in the ability to identify and translate voice, it falls behind compared to larger models.
Unlike Gemini 1.5 Pro is the Moe model (Mixture of Experts), Gemini 1.5 Flash is a dense model, filtered online from a larger 1.5 Pro model to improve quality. In terms of speed, the Flash model surpasses all the smaller models available, including Claude 3 Haiku, running on custom TPU of Google.
And its price is so low that it is unbelievable. Gemini 1.5 Flash costs $ 0.35 for inputs and $ 0.53 for output to process 128k tokens, 0.70 USD and 1.05 USD for 1 million tokens. It is much cheaper than Llama 3 70B, Mistral Medium, GPT-3.5 Turbo and of course larger models.
If you are a developer and need multimodal deduction with larger context window at a cheap price, you should definitely look through the flash model. This is a free way to try Gemini 1.5 Flash.
How to use Flash Gemini 1.5 for free
1. Please visit Aistudio.google.com And log in with your Google account. There is no waiting list to use the Flash model.
2. Next, choose the model “Gemini 1.5 Flash” In the Drop-Down menu.
3. Now, you can start chatting with the flash model. You can also upload to photos, videos, audio clips, files and folders.
The first impression of Gemini 1.5 Flash
Although Gemini 1.5 Flash is not a modern model, its advantage is dizzying speed, efficiency and low cost. It is ranked under Gemini 1.5 Pro in terms of capabilities and other models from Openai and Anthropic. However, the author has tried some theoretical suggestions used to compare chatgpt 4o and Gemini 1.5 Pro.
It can only create a correct answer among 5 questions. It may not be very intelligent in terms of common reasoning, but for other applications that require multimodal ability and large context window, it can be suitable for your use. In addition, the Gemini model is very good in creative tasks that can bring value to developers and users.
Put simply, no one model is fast, effective, multi -method and has a large context window with almost perfect ability. Above all, it is extremely cheap.
What is your opinion on Google's latest Flash model? Share your comments in the comment section below!
