AIMarketCap | AI Trends & Prompts

Higgsfield launches Shorts Studio Shorts Studio is a tool for creating attention-grabbing short-form videos that hook viewers from the very first second. It runs on Google’s Gemini Omni Flash model. Upload any video, and the tool automatically adapts it for short-form platforms like TikTok, Reels, and Shorts. It optimizes the editing, pacing, and rhythm to maximize viewer retention and keep people from scrolling away. Available through the Higgsfield platform, MCP, and Supercomputer. The best AI on AIMarket 🚀

Fable 5 Is Back Anthropic has announced the return of access to Fable 5 after U.S. authorities lifted the restrictions on its use and export. The model had previously been taken offline due to security requirements and access restrictions for foreign users. The suspension affected even Anthropic employees, bringing all work involving Fable 5 to a complete halt. Anthropic has since reached an agreement with U.S. authorities and is bringing the model back online. The company says it has strengthened the system’s security, introduced new mechanisms to filter potentially dangerous requests, and will continue working closely with regulators on safety-related issues. Access will be restored gradually, with additional verification measures and restrictions on sensitive use cases. In practice, this means that in many cases requests will continue to be routed to Opus 4.8 until Fable 5’s safety systems and request handling are fully calibrated. So it’s not all smooth sailing just yet. The best AI on AIMarket 🚀

Gamma has arrived in ChatGPT Gamma now works directly inside ChatGPT. Just type your prompt in the chat, and you’ll get a polished presentation that you can further refine in Gamma. With Gamma in ChatGPT, you can turn a technical report into a concise executive summary, create a marketing brief for a product or service, or build a presentation with a few slides for a specific occasion—like a team pitch. The best AI on AIMarket 🚀

Microsoft turns Excel Copilot into a financial analyst Microsoft has upgraded Copilot in Excel for finance teams, introducing three major improvements: reusable skills, direct access to financial data, and greater transparency. Skills are predefined workflows for recurring tasks such as company valuation, quarter-end close, or board reporting. Instead of generating a new approach every time, Copilot follows a structured process. Users can choose built-in skills or create their own using markdown files stored in OneDrive. Copilot can now pull data directly from six financial sources, including CB Insights, PitchBook, Daloopa, FactSet, Morningstar, and S&P Global, eliminating the need for manual copy-pasting. Transparency has also improved. Before making changes, Copilot now shows exactly which cells, formulas, and worksheets it plans to modify. All edits are logged and clearly marked as AI-generated. The best AI on AIMarket 🚀

Claude handled a robot dog 20 times faster than humans Anthropic tested Claude Opus 4.7 on a quadruped robot, assigning it technical tasks previously performed by human teams: connecting to cameras and sensors, writing control software, and setting up object detection. On the four tasks completed by both human teams, Claude finished in just 9 minutes and 35 seconds. A team with access to Claude took 181 minutes, while a team without AI needed 361 minutes. That makes Claude 37 times faster than humans working without AI and nearly 19 times faster than those using AI. Claude also wrote ten times less code than the Claude-assisted team in an earlier experiment, and most of its code worked on the first try. The main challenge remains autonomous ball retrieval. Humans can master this after several training sessions, but Claude still struggles because the task requires continuous real-time corrections. The best AI on AIMarket 🚀

OpenAI has released GPT-5.6 GPT-5.6 is a new family of three models. Sol is the most powerful. Terra offers a balance between capabilities and cost, operating at half the price of GPT-5.5. Luna is the fastest and most affordable model. Sol achieves the best results among all OpenAI models on benchmarks for coding, biology, and cybersecurity. On Terminal-Bench 2.1, which evaluates command-line performance, Sol sets a new state-of-the-art record. On ExploitBench, a cybersecurity benchmark, Sol delivers results comparable to Mythos Preview while using three times fewer tokens. Pricing per 1 million tokens: Sol costs $5 for input tokens and $30 for output tokens. Terra costs $2.50 and $15, respectively. Luna costs $1 and $6. Access is currently limited. OpenAI shared the models with the U.S. government in advance and, at its request, is initially rolling out GPT-5.6 to a small group of trusted partners. Broader availability through ChatGPT and the API is planned in the coming weeks.

Midjourney has updated two tools at once, making it faster to work with styles First update: a new --preview parameter has been introduced. Add it to your prompt, and you’ll see how the image will look in V8.2 without running a full generation. It’s a quick way to check your settings before spending time on the final render. Second update: batch image processing now works together with the --sref parameter. Previously, cycling through styles in batch mode was time-consuming. After the update, the process is now 24 times faster. The best AI on AIMarket 🚀

Claude now lives directly inside Notion Notion has introduced support for Claude agents in beta. Users can now work with Claude without leaving their workspace. Here’s how it works: a new Agents section has appeared in the Notion sidebar. You can choose Claude, start from a prebuilt template or from scratch, and set up instructions and triggers. The agent can only access the pages and databases it has been granted permission to use. The agent can answer questions in chat, work with documents and task boards, and create or edit content. What it can’t do: browse the web or call other agents. Perhaps the coolest part is that agent usage is billed through Notion credits. There’s no need for your own Anthropic account — the API keys and infrastructure are managed entirely by Notion. The feature is currently available only on Business and Enterprise plans.

🚀 ByteDance has unveiled Seedance 2.5 The company introduced Seedance 2.5, a new AI video generation model. Its biggest feature is the ability to generate videos up to 30 seconds long in a single pass, without stitching together short clips. What’s new:

▪️ Native 4K video generation; ▪️ Up to 50 image, video, and audio references in a single prompt; ▪️ Edit specific parts of a video without regenerating the entire clip; ▪️ Support for 3D models for precise scene and camera control.

The public release of Seedance 2.5 is expected in early July. Looks like Sora and Veo have a serious new competitor. 🎬🔥 The best AI on AIMarket 🚀

🚨 Claude is experiencing a major global outage If Claude isn’t opening today (June 23), responses are loading slowly, or the service is completely unavailable the issue is not on your side. Users around the world are reporting widespread problems with the AI service. More than 8,000 reports were submitted within just a few minutes. Here’s what we know so far:

▪️ Claude may fail to open; ▪️ The app may not load for some users; ▪️ Errors may occur when sending prompts or generating responses; ▪️ The developers are aware of the issue and are working on a fix.

Some users are also reporting issues with ChatGPT. If AI services are unstable today, the best option is to wait for an official fix. Is Claude working for you right now? 👇

Happy Horse 1.1 is out, but it’s not exactly a revolution The model can generate videos with built-in voiceover in the native language right from the start, while keeping lip movements synchronized with the spoken words. It works across multiple languages, without the need for separate dubbing or lip-sync workflows. Three video creation modes are available: from text, from an image, or from an existing video used as a starting point. In other words, you can begin with a single text prompt, a single photo, or take an existing clip and transform it into something new. For greater control, you can upload up to nine reference images. The model uses them to maintain the desired visual style, character appearance, and environment throughout the video. Two output resolutions are available: 720p and 1080p. This lets you choose between faster generation times and cleaner, publication-ready visuals. You can try it right now on the fal platform⁠.

Codex now works with any models It looks like OpenAI has become truly open. Codex now includes support for working with local models. You can connect it for free to third-party platforms such as Ollama or LM Studio. To enable this mode, simply use the oss command when launching Codex. If no specific application is specified, Codex will automatically connect to the default platform. Developers have set Ollama as the default option, but this can easily be changed in the configuration file via the oss_provider setting. The new feature makes it possible to completely move away from the cloud. You can run any compatible neural networks on your own hardware and avoid paying for third-party subscriptions. The best AI on AIMarket 🚀

Perplexity has taught its AI agent to learn from its own mistakes Perplexity has introduced Brain, a new memory system for its Computer agent. Most AI systems remember you — your preferences and the way you work. Brain works differently: instead of remembering the user, it remembers what the agent did, what worked, what failed, and what corrections had to be made. While you’re asleep, Brain reviews the agent’s entire work history overnight, extracts lessons from it, and updates its working context. The next day, the agent starts with this accumulated experience and avoids repeating past mistakes. The more tasks it completes, the smarter it becomes. According to Perplexity’s internal tests, answer accuracy on recurring tasks increased by 25%, retrieval quality of relevant information improved by 16%, and operational costs on tasks with historical context dropped by 13%. Brain is currently available in testing mode for Perplexity Max and Enterprise Max subscribers. The best AI on AIMarket 🚀

Old Money Style Transform your photo into the Old Money aesthetic:

Transform the photo into an Old Money style. Add the atmosphere of old European aristocracy, exclusive tennis clubs, yachting holidays, luxury resorts, and timeless elegance. Use natural lighting, refined color palettes, premium fabrics, and the aesthetic of Ralph Lauren campaigns and luxury lifestyle magazines. Keep the person completely recognizable. Do not alter the face, body shape, age, clothing, or pose.

The best AI on AIMarket 🚀

Dopamine Art (this style gained 1M+ reach and 25K likes on Threads) Create pictures using your appearance with this prompt:

Please transform the entire image into a single Decorative Folk Flat Illustration with Doodle elements. Use a bold and playful color palette, completely different from the original image. Simplify all details into clean, flat shapes with a handmade, slightly imperfect feel, as if drawn on a sheet of white paper. The overall style should look cute, childlike, and whimsical

The best AI on AIMarket 🚀

Create animations in just 2 clicks A new open-source project called Lottie has been released, making animation workflows much simpler. Instead of dealing with complex interfaces and endless exports, you can work with animations as part of your code and instantly get production-ready results. Perfect for web apps, UI animations, landing pages, and product interfaces where speed and lightweight performance matter. The best AI on AIMarket 🚀

Artistic Portrait Create adorable pictures using your appearance with this prompt:

Vibrant impressionist watercolor painting. Wearing a sun-hat and a textured linen suit, composed of thick watercolor impressionist brushstrokes. Dappled sunlight filters through trees, capturing a relaxed and candid expression.

The best AI on AIMarket 🚀

MiniMax has launched Hub, a local workstation where AI agents handle creative work for you MiniMax Hub is a workspace where a team of AI agents takes care of the routine tasks while you focus on direction and decisions. The focus has shifted from coding to creativity: agents do the groundwork, while you stay in control. Dedicated agents handle each stage of the process: research, scriptwriting, image generation, music creation, and final editing. The result is a complete workflow from idea to finished video in a single place. It’s built for serious projects. You can run multiple projects in parallel, generate content in batches, and work on an infinite canvas that keeps everything visible at once. Under the hood are powerful AI models, enhanced with custom skills and carefully selected toolsets. Hub works directly with your computer, connecting local files and applications while enabling instant import and export of assets. The best AI on AIMarket 🚀

Meet @AIMarketCap_bot— your AI assistant right inside Telegram What can it do? ✨ Answer your questions on any topic 🎨 Generate AI images 📸 Edit and enhance your photos 📄 Analyze documents, PDFs, presentations, and Word files 💬 Ask questions about uploaded files and get instant answers To get started, every user receives 2 free generations. Try it now: @AIMarketCap_bot More features and updates coming soon 🚀

US Government Blocks Anthropic’s Newest AI Models Anthropic has completely disabled access to its most advanced AI models, Fable 5 and Mythos 5, for all users worldwide. The move follows an emergency order from US authorities under export control regulations. The government reportedly required Anthropic to immediately restrict access to these technologies for all foreign nationals, both inside and outside the United States, including the company’s own international employees. Anthropic disagreed with the decision, stating that the models feature robust multilayer safety systems and that the sudden restriction undermines transparency and fairness. However, the company confirmed it is complying with the legal order. Other Anthropic AI models remain fully operational. The company is now working with government agencies to resolve the issue and restore access as quickly as possible. The best AI on AIMarket 🚀