Android version of ChatGPT App is the first to land in four countries
Android version of ChatGPT App is the first to land in four countries
The Android version of ChatGPT App is the first to land in four countries including the United States and India
Two months after Apple users tried it out, Android users can finally use ChatGPT on their phones.
On Tuesday, July 25th, Eastern Time, OpenAI announced that users in the United States, India, Bangladesh, and Brazil can now download the Android version of the ChatGPT app on Google Play, and provided relevant links.
OpenAI said it plans to roll out the app to more countries next week.
When the iOS version of the ChatGPT app was released in May, OpenAI said it would be coming to Android “soon.” Consistent with OpenAI’s preview last week, the Android version of ChatGPT will arrive as scheduled this week.
ChatGPT’s App can synchronize the user’s history on different devices and support voice input.
Users can use the App to ask ChatGPT questions, get instant answers, and also get ChatGPT’s suggestions, let ChatGPT write the text of emails or PPT presentation files.
The arrival of the Android version coincides with the end of the explosive growth of ChatGPT users, with both downloads and visits declining.
Wall Street News previously mentioned that SensorTower data showed that in June this year, the installation volume of ChatGPT and Microsoft Bing App fell by 38%, while SimilarWeb data showed that the global desktop and mobile traffic of the ChatGPT website in June decreased by 9.7% compared with May. , unique visits to the site fell by 5.7%, and the time spent on the site per user decreased by an average of 8.5%.
Moreover, the new version of GPT faces more “intelligence-reducing” doubts.
Some users pointed out last month that the large model ceiling GPT-4 has become stupid.
Recently, there are papers demonstrating that GPT-4’s mathematical ability has avalanche, and its coding ability has also deteriorated.
Some experts in the field of language models speculate that this is because OpenAI is creating multiple small GPT-4 models that behave similarly to large models but have lower operating costs, using this mixed expert model (MOE) to quietly engage in “cost reduction and efficiency increase “.
OpenAI has responded that since the release of GPT-4 on March 14, the ontology of the large model has been static, and there is no large amount of external data polluting the model.
However, it admits that the large model itself is unstable, so for similar prompt words , the large model has inconsistencies in the answers.
OpenAI said last week that it will continue to update the API.
In addition, AI competition is in full swing, and OpenAI is also facing emerging opponents.
Last week, OpenAI’s big “golden master” Microsoft “split” Meta and announced that it will cooperate with Meta to release the commercial version of the open source AI model Llama, providing enterprises with alternative products of OpenAI and Google models.
