Google is celebrating Translate’s 20th birthday by launching pronunciation practice, which the company says is one of the most requested features for the product. The feature is only rolling out on Android at the moment for English, Spanish and Hindi in the US and India. If it’s available for you, you’ll see a button at the bottom of the app that says “Practice,” which gives you the option to either “pronounce” what you’ve translated or to “listen” to how it’s actually pronounced by native speakers.

If you choose the “pronounce” option, Translate will listen to you speak and then use artificial intelligence to analyze how you said the words to provide instance feedback. It will then show you a phonetic spelling of how specific words should be pronounced. In the example Google provided, for instance, the speaker pronounced the Spanish word for juice as “jugo” with the English “j” sound instead of with the Spanish “j” sound. So, Translate spells it out as “HU-go” in its pronunciation suggestion.

Google said around third of users on mobile use Translate to practice speaking and listening in order to be able to hold real-world conversations, making this new feature a very useful addition. The company also revealed other stats about the app. Apparently, it now supports over 250 languages, including some endangered and indigenous ones, and has over 1 billion monthly user who have been translating over 1 trillion words every month.

DeepSeek has released its latest AI models, the V4 Pro and Flash versions, a bit over a year after it went viral and became the top rated free app on Apple's App Store in the US. “Welcome to the era of cost-effective 1 million context length,” DeepSeek said in its announcement. Context length is what you call the maximum number of tokens that an AI model can remember, so the bigger it is, the more coherent and consistent an AI is when it comes to extended conversations. OpenAI’s recently announced GPT‑5.5 has a context window ranging from 400,000 to 1 million, for instance.

The new model is still open-source, allowing users to download its code and modify it if they want. DeepSeek says V4 Pro has enhanced agentic capabilities and claims that it rivals top closed-source models when it comes to reasoning. It also says that it trails only Gemini-3.1-Pro in rich world knowledge. Meanwhile, V4 Flash isn’t quite as powerful as the V4 Pro, but it has faster response times. Still, its reasoning abilities closely approach V4 Pro, DeepSeek says, and it performs on par with with the Pro version on simple Agent tasks.

Shortly after DeepSeek topped the App Store charts, it was banned for use by US federal agencies and on government-owned devices. Authorities believed it was a national security risk and posed a threat to US AI stocks. South Korea also paused downloads of its app over privacy concerns.