Top AI Innovations This Month: New AI Tools

The world of artificial intelligence is evolving at an incredible pace, with new solutions emerging every month that are transforming how we work, create, and tackle daily tasks. In this article, we’ve compiled the top AI innovations this month, highlighting the new AI tools you might have missed. We’ve categorized these advancements to help you easily find what you need, from video and audio generation to coding and design creation.

AI Tools for Video and Animation

Video generation continues to impress with its capabilities. This month has seen several powerful tools for creating and editing video content.

•LongCat-Video-Avatar 1.5 — An updated version of the model that has learned to create realistic talking characters. It ensures stable video generation with audio and supports multi-character interaction within the frame.

•AI Video Upscaler by ByteDance — A tool for enhancing video quality. It allows for increasing the resolution and clarity of video materials using artificial intelligence algorithms.

•Grok Imagine Video 1.5 by xAI — An updated video generation model from Elon Musk’s company, offering improved quality and detail in generated clips.

•Seedance 2.0 and 2.5 — Video models from Dreamina. Version 2.0 has learned to build camera movement along a drawn line and generate 4K video, while version 2.5 gained the ability to create 30-second clips. A budget version, Seedance 2.0 Mini, is also available.

•Happy Horse 1.1 — An open-source video generation model that now supports creating videos with voiceovers and lip-syncing.

•Luma Ray 3.2 — This model now supports HDR and EXR formats, significantly expanding its professional use cases, though its pricing policy might be a surprise.

•Cutback Selects — A service that automates the first stage of video editing, significantly speeding up content creation.

•LTX Trainer — The tool has been updated for training LoRA models on video, allowing for the creation of custom styles and effects.

•HeyGen frame.md — A new tool for creating branded AI videos, which also addresses the issue of AI avatar consistency.

•Google Vids — Google’s service now offers free access to AI avatars for creating video presentations.

AI for Programming and Development

Developers have gained a whole arsenal of new assistants that automate code writing, bug finding, and architecture creation.

•Codex by OpenAI — Received a massive update. It can now see the screen and work remotely on macOS, operate with Windows, accumulate limit resets, repeat tasks after a single demonstration, and has gained skill sets for business, teamwork, and OSS Mode. Additionally, OpenAI simplifies iOS app development with Codex.

•Grok Build by xAI — A new AI agent for working with code, assisting developers in creating and optimizing software.

•Claude Code by Anthropic — Received a plugin for checking code vulnerabilities, as well as Artifacts support for more convenient work with generation results.

•Qwen 3.7-Max — A new model that surpassed Claude and GPT in self-improving agent tests, demonstrating high efficiency in solving complex tasks.

•Cursor — The popular code editor has taught its agents to work more autonomously. Interestingly, rumors have emerged about SpaceX acquiring Cursor for $60 billion.

•Kimi-K2.7-Code — A new flagship model from Moonshot AI, specifically designed for programming tasks.

•Serge — A tool that adds an AI layer on top of GitHub Code Review, automating the code review process.

•MiMo Code — A new tool that aims to solve the main problem of AI agents for programming by offering new approaches to code generation.

•Google AI Studio — Now allows generating Android applications directly through its interface.

AI for Design, Images, and 3D

The visual content sphere has been enriched with tools that simplify interface creation, image generation, and 3D graphics work.

•HiDream — Released a universal model for working with images, combining various editing and generation functions.

•Krea 2 — Received a new Moodboard gallery with ready-made references and automatic style selection. Sliders have also been implemented for changing images without rewriting the prompt. Krea 2 Raw and Turbo versions are now open-source, with Turbo generating images in seconds.

•Miora — A new service that combines image, video, and 3D generation in one convenient interface.

•Ideogram 4.0 — The new version of the popular text-to-image generator is now available open-source.

•Reve 2.0 — An updated image generator offering improved quality and new styles.

•Midjourney — Accelerated its Style Reference feature, allowing for faster application of selected styles to new generations.

•FLUX VTO — An innovative tool that allows trying on clothes via AI, creating realistic visualizations.

•Figma — Enabled code editing through its interface and is transforming into a platform for design, code, and AI. Figma Bot also appeared, allowing any website to be converted into an editable Figma file.

•Claude Design — Has learned to automatically adhere to a company’s brand style when creating designs.

•Genspark — Updated AI Slides to version 5.0 and combined design and code generation into one tool.

•400+ DESIGN.md — A large collection of files for creating more beautiful interfaces through AI.

•OpenPencil — Offers a local alternative to AI design editors.

•PlayCanvas 2.20 — Enhanced its work with Gaussian Splats and WebGPU for creating interactive 3D content.

•Diffusion Studio — Opens a library for working with Lottie animations via code.

Audio, Music, and Voice AI

Sound generation is reaching new heights: from creating full-fledged musical tracks to realistic dubbing and voice translation.

•ElevenLabs — The company introduced a range of updates: Music V2 music generator, Dubbing v2 AI dubbing, Flows Agent for creating AI pipelines, Ads Engine for translating ads into 50+ languages, and also entered the AI avatar market.

•Qwen3.5-LiveTranslate — The model has learned to translate voice in real-time while preserving the speaker’s original intonation.

•Magenta RealTime 2 — Generates music without pauses or renders, providing a continuous audio stream.

•Suno Advanced Split — A new feature for working with stems, allowing generated tracks to be separated into individual tracks (vocals, instruments).

•SeedAudio 1.0 — A new model from ByteDance that combines voice, music, and sound effect generation.

•Google Translate — Received an update that allows it to almost eliminate pauses during conversation, ensuring more natural communication.

AI Agents, Platforms, and Large Language Models (LLMs)

The development of autonomous agents and new language models continues to shape the future of human-computer interaction.

•Command A+ by Cohere — An open 218B-model specifically designed for agentic tasks.

•Claude Opus 4.8 and Fable 5 — New flagship models from Anthropic. The company also opened a set of working plugins for Claude and demonstrated how to build startups using its models. Additionally, Claude Tag appeared, transforming Claude into a team member in Slack.

•GPT-5.6 and GPT-5.5 Instant — OpenAI introduced a new family of GPT-5.6 models, and GPT-5.5 Instant received improved medical responses. ChatGPT can now update its memory about the user and send emails.

•Gemini by Google — Gained the ability to create its own digital twin, a personal learning course generator, and version 3.5 Flash received built-in Computer Use support. Google also released an official prompt guide for Gemini Omni.

•Microsoft 365 Copilot and Autopilots — Microsoft redesigned its Copilot, released AI Engineering Coach and SkillOpt, and introduced a new generation of AI agents called Autopilots and local AI for Windows.

•Perplexity — Opened the Bumblebee code for testing AI tools, introduced Search as Code for AI agents, Perplexity Brain for training agents on their own mistakes, and the Enterprise version received a solution for legal teams. The company also plans to move part of its AI computations from the cloud to users’ PCs.

•Odysseus by PewDiePie — The famous blogger introduced a platform for deploying a local AI workspace.

•Hermes Agent — The AI agent can now be installed as a regular application or assembled from a minimal configuration.

•MiniMax — Released the open-source M3 model and launched Hub — a platform where AI agents create content for you.

•Gemma 4 12B — Now supports audio and local deployment.

•GLM-5.2 — Received a 1 million token context, and Unsloth prepared GGUF versions for local deployment.

•NotebookLM by Google — Is transforming into an autonomous AI researcher.

•Kimi Work — Launches up to 300 AI agents to solve a single task.

•AgentBase — A platform that transforms corporate data into ready-to-use working systems.

•Fugu by Sakana AI — A new model that aims to displace Fable 5 from the AI agent market.

•HappyOyster by Alibaba — The company released an open-source world model.

Other Interesting Tools and News

•Muranyi 3 — A new model that has learned to create full-fledged games through text descriptions.

•100 open-source repositories — A large collection of useful tools gathered in one place.

•Manus — Transferred the Projects feature to its mobile application.

•Ask YouTube — The platform is testing a new type of AI search for more convenient video navigation.

•Higgsfield — Integrated its AI tools into Adobe Premiere Pro and Figma, released a plugin for Photoshop, and is also entering the AI game development market.

•CapCut Design Studio 2.0 — The popular editor received a major update to its design tools.

•Dreamina Octo — A new AI assistant has appeared in the service.

•Cognition Windsurf — Has transformed into a platform for managing AI agents.

•LM Studio — Transferred the ability to run local AI to iPhone and iPad.

•Dreambeans by Google — A new service that transforms your data into personalized AI stories.

•Apple Intelligence — The company rebooted Siri and integrated Gemini to expand its assistant’s capabilities.

•Google Flow — Received content generation based on real addresses.

•Lovable — Adds visual project editing for easier development.

•AMD — Is bringing the launch of large AI models closer to regular PCs by optimizing hardware.

•OCR 4 by Mistral — A new model that effectively converts documents into structured data.

This month has brought many innovations that make artificial intelligence even more accessible and useful in various fields. Stay tuned for updates so you don’t miss the most interesting AI tools and neural networks!