Claude Code expands its reach to terminal, IDE, desktop, and browser, introducing a /desktop command. Cursor introduces long-running autonomous agents capable…
Claude Code
Claude Code is now available across your terminal, IDE, desktop app, and browser.
You can customize Claude Code for your specific workflow through its settings.
A new /desktop command allows you to hand off a terminal session to the Desktop app for visual diff review.
Cursor
Cursor can now work autonomously over longer periods to complete more complex tasks without constant human intervention.
These long-running agents have successfully completed work previously too difficult for regular agents, resulting in more complete code changes.
The long-running agent feature is now available for users on Ultra, Teams, and Enterprise plans.
Perplexity
Perplexity now intelligently selects between images, videos, or both based on your search query type.
You can enable, disable, and override the chosen media types as needed for more control over your search results.
Smart media selection : Intelligently chooses between images, videos, or both based on query type
DeepSeek
DeepSeek-V3.2, the official successor to V3.2-Exp, is now live and accessible on the App, Web, and API.
DeepSeek-V3.2-Speciale: Pushing the boundaries of reasoning capabilities. API-only for now.
Introduces a new massive agent training data synthesis method covering 1,800+ environments & 85k+ complex instructions.
Claude
More than a billion people in India speak one of over a dozen officially recognized languages, but AI models continue to perform better in English than they do in other languages.
Now, Anthropic is working with Karya and the Collective Intelligence Project to build evaluations testing performance on locally relevant tasks across domains like agriculture and.
ChatGPT
In a new preprint, GPT‑5.2 proposed a formula for a gluon amplitude later proved by an internal OpenAI model and verified by the authors.
We’ve published a new preprint showing that a type of particle interaction many physicists expected would not occur can in fact arise under specific conditions.
The preprint shows that this conclusion is too strong.
OpenAI Codex
Underneath this is a complex system that fuses limits, real‑time usage tracking, and credit balances in a single access model.
For Codex and Sora, neither was sufficient on its own.
When a user hits a limit and has credits available, the system must know immediately .
Gemini
Setting a new standard (48.4%, without tools) on Humanity’s Last Exam, a benchmark designed to test the limits of modern frontier models
Our most specialized reasoning mode is now updated to solve modern science, research and engineering challenges.
We updated Gemini 3 Deep Think in close partnership with scientists and researchers to tackle tough research challenges — where problems often lack clear guardrails or a single co.