This week in AI: OpenAI’s o3, Anthropic’s agentic Research, and a few updates from Chinese vendors
A recap of the latest AI tools released by OpenAI, Anthropic, and a few other companies.
Cheng Xin/Getty Images
A recap of the latest AI tools released by OpenAI, Anthropic, and a few other companies.
With AI's rapid advancements, major announcements regarding new AI models and tools occur every week. This week was no different, with major companies like OpenAI, Anthropic, and some Chinese vendors releasing new developments.
An exciting update came from the Chinese short video company Kuaishou Technology. It has just released a new video generator, Kling AI 2.0, calling it “the most powerful video generator.”
According to third-party firm Artificial Analysis, Kuaishou’s previous video generator was ranked as the world’s top image-to-video model and the second-best text-to-video model.
The new release reportedly offers improved prompt comprehension, better adherence to instructions, and enhanced image and motion quality, bringing more competition to OpenAI’s Sora.
OpenAI has released new reasoning models, o3 and o4-mini, which can now autonomously use and combine all available tools within ChatGPT, as well as Codex CLI, a new open-source coding agent that runs locally on a user's terminal.
According to OpenAI, o3 excels in coding, math, science, and visual perception, and makes 20% fewer major errors than o1 in tasks such as programming, business consulting, and creative ideation.
The o4-mini, a smaller model optimized for fast reasoning, also features notable improvements and is currently OpenAI’s best-performing model on AIME standardized tests.
Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date.
undefined OpenAI (@OpenAI) April 16, 2025
For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation. pic.twitter.com/rDaqV0x0wE
This week, Anthropic’s Claude made a move to catch up with major competitors like ChatGPT, Gemini, and leading Chinese chatbots. Anthropic released an agentic research feature capable of conducting multi-step research tasks.
Additionally, Claude now integrates with Gmail and Google Calendar, enabling more seamless productivity.
Today we’re launching Research, alongside a new Google Workspace integration.
undefined Anthropic (@AnthropicAI) April 15, 2025
Claude now brings together information from your work and the web. pic.twitter.com/FfHXJLCJf7
The online design and publishing platform Canva also unveiled a new tool, Visual Suite 2.0. New features, such as Canva Sheets and Canva Code, allow users to generate tools and custom widgets from prompts, turn data into visualizations, and more.
Meanwhile, ByteDance, the parent company of TikTok, launched a new video generation model. While it has only 7 billion parameters, the model supports synchronized audio-video generation, long-shot storytelling, and real-time high-resolution output.
Finally, Elon Musk’s X has released a beta version of a new memory feature that can remember all user conversations. Similar capabilities already exist on platforms like ChatGPT, Gemini, Meta AI, and Copilot.