# AI Digest for October 2026 — New Models, Personal Agents, and Work Tools

We’ve gathered the top news from the world of artificial intelligence for October — from models for automating office tasks and creating virtual worlds to AI agents for business and free subscriptions for students in Uzbekistan.

[**Google introduced the new AI model Gemini 4 Argon**](<https://the-tech.kz/google-predstavila-novuyu-ii-model-gemini-4-argon/>)Google introduced Gemini 4 Argon — a model designed for complex tasks in programming, corporate work, and cybersecurity. It features a context window of one million tokens, enabling it to process large volumes of information and work longer on complex tasks without restarting the process. Inside Google, the model already assists engineers in writing code, exploring data, and optimizing systems.

[**Students in Uzbekistan will be able to use Google Gemini AI Plus for free**](<https://the-tech.kz/studenty-v-uzbekistane-smogut-polzovatsya-google-gemini-ai-plus-besplatno/>)Students in Uzbekistan can receive a one-year subscription to Google Gemini AI Plus, valued at $60, for free. To do this, they must verify their student status through the aileaders.uz platform. This opportunity arose after the international verification service SheerID added Uzbekistan to its list of supported countries. The initiative was implemented in collaboration between the Ministry of Digital Technologies of Uzbekistan and Google.

[**Meta introduced new AI models Muse Wonder for personal AI agents**](<https://the-tech.kz/meta-predstavila-novye-ii-modeli-muse-wonder-dlya-personalnyh-ii-agentov/>)Meta introduced Muse Wonder and Muse Wonder Full — models for personal and professional AI agents. They are capable of remembering previous conversations, retaining important information about the user, and working longer on complex tasks. The models can also analyze documents, write code, and handle information in the fields of law, finance, and health.

[**Now you can run your own AI-generated apps on Google Play**](<https://the-tech.kz/teper-v-google-play-mozhno-zapuskat-sobstvennye-prilozheniya-napisannye-ii/>)Google has allowed users to install personal applications created using AI in Google AI Studio on Android devices. A developer account is not required; a regular Google account is sufficient. The user describes the app idea, and the AI generates the code. The finished result can be installed via Google Play, but it will be available only to the creator and will not appear in the public catalog.  

[**Google launched the Gemini AI agent for corporate tasks**](<https://the-tech.kz/google-zapustila-ii-agenta-gemini-dlya-vypolneniya-korporativnyh-zadach/>)Google introduced a corporate AI agent, Gemini, which can work with user goals rather than just executing individual commands. It can independently select suitable models and connect to workplace services, including Google Workspace, Microsoft 365, Slack, Jira, Confluence, and corporate databases. Meanwhile, companies will be able to manually select models, including using Anthropic solutions.

[**Perplexity launched an AI agent for Mac with local data processing**](<https://the-tech.kz/perplexity-zapustila-ii-agenta-dlya-mac-s-lokalnoj-obrabotkoj-dannyh/>)Perplexity added a local mode for the Computer AI agent on Mac computers with Apple silicon processors. It can now analyze files and interact with supported applications directly on the device without sending personal data to the cloud. For example, the agent can review tasks in notes, check colleagues' messages, and match expenses in spreadsheets to prepare a report.  

[**Google DeepMind introduced a model for creating interactive 3D worlds**](<https://the-tech.kz/google-deepmind-predstavila-model-dlya-sozdaniya-interaktivnyh-3d-mirov/>)Google DeepMind introduced Genie 4 — a model that creates interactive virtual worlds in real time. Users can navigate through them, control characters, and change the environment using text commands. The model generates graphics, physics, sound, and dialogue, and also allows changing the weather, adding objects, and moving the camera directly during the simulation.

[**OpenAI released GPT-6.1 Terra for automating office tasks**](<https://the-tech.kz/openai-vypustila-gpt-6-1-terra-dlya-avtomatizaczii-ofisnyh-zadach/>)OpenAI introduced GPT-6.1 Terra — a model that independently performs tasks on a computer and works with various applications. It can draft proposals, fill out forms, analyze technical drawings, and edit videos. GPT-6.1 Terra is available to Plus, Pro, Enterprise, and Edu subscribers in ChatGPT Work and Codex, as well as to developers via the API.

[**Anthropic introduced Claude Duet, a personal AI agent for business**](<https://the-tech.kz/anthropic-predstavila-claude-duet-personalnogo-ii-agenta-dlya-biznesa/>)Anthropic introduced Claude Duet — an agent that tracks work tasks and notifies the user about items requiring attention. It connects to workplace services, generates morning summaries, finds missed emails and messages, prepares responses, checks unpaid invoices, and helps create documents and reports. The agent has access to over 6,000 connectors and plugins from the Claude Marketplace.

[**NVIDIA introduced an AI model for audio transcription and translation**](<https://the-tech.kz/nvidia-predstavila-ii-model-dlya-rasshifrovki-i-perevoda-audio/>)NVIDIA released Nemotron 3.5 Transcribe — an open model that transcribes audio recordings, translates speech, identifies speakers, and creates brief meeting summaries. It supports 40 languages and can perform multiple tasks from a single instruction. Along with the model, the company released an open set of audio recordings, Granary v2, in 48 languages for training and customizing speech recognition systems.
