Perplexity got smarter search, Claude Code made coding easier, DeepSeek launched a new version, Cursor added integrations, and Claude released a super smart up…
Perplexity
Perplexity can now intelligently choose between images, videos, or both to show you based on what you ask.
You can now choose if you want to see images, videos, or both in your search results, and even turn them off if you prefer.
Smart media selection : Intelligently chooses between images, videos, or both based on query type
Claude Code
Claude Code now uses a newer, smarter brain called Sonnet 4.5 by default.
You can now use Claude Code right inside your VS Code app, seeing its changes as they happen in a special side panel.
Claude Code can now save its progress automatically when working on its own, like creating 'checkpoints'.
DeepSeek
A new, improved version of DeepSeek, called V3.2, is now available to use in the app and on the web.
DeepSeek-V3.2: Official successor to V3.2-Exp. Now live on App, Web & API.
DeepSeek-V3.2-Speciale: Pushing the boundaries of reasoning capabilities. API-only for now.
Cursor
Cursor now works with over 30 new apps like Atlassian, Datadog, and GitLab, letting it help you across more of your work.
We've added more than 30 new plugins from partners such as Atlassian, Datadog, GitLab, Glean, Hugging Face, monday.com, and PlanetScale.
Most plugins contain MCPs that cloud agents can use when kicked off manually or triggered automatically through automations .
Claude
A new, super smart version of Claude, called Opus 4.5, is now available for you to use.
You can find Opus 4.5 in the Claude apps and on major cloud platforms.
You can now use Claude in more places, like inside Excel, Chrome, and on your computer.
Grok
Grok is an AI assistant built by xAI.
Chat, create images, write code, and get real-time answers from the web and X.
Gemini
With a new funding commitment for open source security and helpful tools for defenders, we're shifting from identifying open source security threats to solving them.
Billions of people rely on an Internet built on open source software — which is software anyone can use — but that reliance only works if the software beneath it is secure.
Today, as a founding member of the Linux Foundation's Alpha-Omega Project , we’re pledging $12.5 million collectively with Amazon, Anthropic, Microsoft/GitHub and OpenAI to furthe.
OpenAI Codex
But when we built Codex Security, we made a deliberate design choice: we didn’t start by importing a static analysis report and asking the agent to triage it.
This isn’t just a theoretical pattern.
A reasonable reaction is: why not do both? Start with a SAST report, then use the agent to reason deeper.
ChatGPT
GPT‑5.4 mini significantly improves over GPT‑5 mini across coding, reasoning, multimodal understanding, and tool use, while running more than 2x faster.
GPT‑5.4 nano is the smallest, cheapest version of GPT‑5.4 for tasks where speed and cost matter most. It is also a significant upgrade over GPT‑5 nano.
These models are built for the kinds of workloads where latency directly shapes the product experience: coding assistants that need to feel responsive, subagents that quickly comp.