What's new and What's next in AI by TestingCatalog News

12 Dec 2025 · 1 min read

World App adds encrypted chat, group payments, and virtual accounts

What's new? World App launches World Chat for encrypted texts, groups, media, payments and mini apps; virtual accounts, crypto support, and earn rewards roll out on iOS and Android.

12 Dec 2025 · 2 min read

OpenAI launches GPT-5.2 that scores 54% on ARC-AGI-2

OpenAI announces GPT-5.2, offering new models on ChatGPT and API, with advances in coding, spreadsheets, tool use, and expanded context windows.

12 Dec 2025 · 1 min read

Google releases Gemini Deep Research agent on APIs

What's new? Google released its Gemini Deep Research agent via the API with access to Gemini 3 Pro; it gathers online data and produces structured reports with JSON outputs;

12 Dec 2025 · 1 min read

Cursor launches visual editor for web app development in browser

What's new? Cursor has launched a visual editor in the Cursor Browser for React-based projects; it uses drag and drop layout, real time property adjustment and natural language prompts;

11 Dec 2025 · 1 min read

Disney partners with OpenAI to bring Disney characters to Sora

OpenAI and Disney announce a three-year partnership giving Sora and ChatGPT access to iconic Disney IP for AI-generated videos and images starting in 2026.

11 Dec 2025 · 1 min read

Google tests Gemini 3 Flash model on LM Arena

Google is testing Fierce Falcon and Ghost Falcon models on LM Arena, signalling a potential Gemini 3 Flash or Gemini 3 Pro GA launch.

11 Dec 2025 · 1 min read

OpenAI debuts Garlic model on LM Arena for public testing

OpenAI’s new Robin High model is now on LM Arena for benchmarking, outperforming previous contenders and offering a glimpse at next-gen AI progress.

11 Dec 2025 · 1 min read

Google expands Gemini TTS with 24 languages, lifelike voices

What's new? Google announced Gemini 2.5 Flash and Gemini 2.5 Pro TTS preview models via the Gemini API in Google AI Studio; They offer lifelike speech synthesis with granular control over pace, tone, and 24-language support;

10 Dec 2025 · 1 min read

Cursor receives 2.2 update with new Debug Mode and more

What's new? Cursor introduced debug mode that instruments runtime logs to diagnose cross-stack bugs; it offers real-time logging across languages and frameworks to pinpoint issues;

10 Dec 2025 · 2 min read

Google launches fully managed MCP servers for AI agents

Google is rolling out fully managed MCP servers globally, enabling developers to create AI agents with remote access to Maps, BigQuery, and more.

10 Dec 2025 · 2 min read

OpenAI testing new Image-2 models on LM Arena

OpenAI is set to debut new image generation models, Image-2 and Image-2-mini, promising higher detail and colour accuracy for ChatGPT and creative tools.

10 Dec 2025 · 2 min read

Notion tests unreleased GPT-5.2 model as OpenAI launch nears

OpenAI's GPT-5.2 model is rumored for release soon, with leaks from Notion hinting at early testing and speculation around a potential launch date.

9 Dec 2025 · 2 min read

Mistral AI launches Devstral 2 open-source coding models

What's new? devstral 2 and small 2 are open-source coding models with 123b and 24b params and a 256k context; mistral vibe cli is a cli for code automation and multi-file orchestration;

9 Dec 2025 · 1 min read

Claude users on Team plans can now tag Claude Code on Slack

What's new? Anthropic integration lets teams delegate coding tasks from Slack using Claude Code; beta release triggers coding sessions from slack chat context and posts session updates;

9 Dec 2025 · 1 min read

Instacart launches instant grocery checkout inside ChatGPT

OpenAI and Instacart launch an in-chat app for meal planning and grocery orders in ChatGPT, with built-in Instant Checkout powered by Stripe payments.

8 Dec 2025 · 2 min read

Anthropic works on Agent Mode for Claude and new avatars

Anthropic is testing a dedicated agent mode for Claude, with a home screen toggle and structured task flows, plus a new pixel avatar feature in settings.

8 Dec 2025 · 2 min read

Z AI releases GLM-4.6V and GLM-4.6V-Flash open models

What's new? Z AI open-sourced its GLM-4.6V series for cloud and local setups; models handle 128k tokens and multimodal input via HuggingFace, ModelScope and z.ai;

7 Dec 2025 · 2 min read

Exclusive: Google set to launch Gemini Nano Banana 2 Flash

Google is preparing to launch Nano Banana 2 Flash within Gemini, offering similar performance to the Pro model at lower costs for broader accessibility.

Latest AI News & Updates