Anycodera Open source web app development environment Developed by hugging the face with the ML growth Ahsen Khaliq ((@_akhaliq on x), began hugging facial areas.
The tool that’s now available For all users of the AI -Code -Sharing -Repository -Maumung faceIntegrates live previews, multimodal inputs and one-click deprivation alles in a hosted environment that allows indie creators without much technical know-how or those that work on behalf of shoppers or large firms to quickly start “vibe-coding” wet apps with the support of face-hosted AI models.
It subsequently also acts as a substitute for services similar to Adorablewhich also enable users to enter in easy English and to encode apps without formal programming knowledge.
Free mood coding, which is driven by Kimi K2
Khaliq built AnyCoder as a private project inside the hug -based ecosystem and as “certainly one of the primary Vibe coding apps” to support Moonshots powerful but small and efficient Kimi K2 model that was launched last week.
With the predominant functionality of anycoder, users can enter plain text descriptions to generate HTML, CSS and JavaScript. These are displayed in a live preview area and might be edited or provided directly. It also comprises sample templates for Todo apps, dashboards, calculators and more.
Any code fully built the open source python development environment environment of Hugging Face, and enables users to explain applications in easy English or upload images and immediately generate working front code.
Khaliq built AnyCoder as a private project inside the hug -based ecosystem.
In a direct message discussion with this Venturebeat journalist, he described it as a “free Open -Source Vibe coding -app”.
However, he also noted that several open source models are supported and users can change between them with a dropdown menu within the control side bar within the left area, including:
- Moonshot Kimi-K2
- Deepseek V3
- Deepseek R1
- Baidus Ernie-4.5-VL
- Minimax M1
- Alibabas Qwen3-235b-A22b
- SMOLLM3-3B
- GLM-4.1V-9b considering
Code from UI images, Web search integration and OCR support
With the Ernie 4.5 VL model, AnyCoder supports the multimodal generation. Users can ui design screenshots or models can upload and generate functional frontend code of you. He is helpful for designers or teams who work visually.
AnyCoder comprises a web site redesign tool that extracts content from every public website and initiates it again with a more modern layout. Scary content similar to side structure, meta information and pictures are used to create a new edition that’s optionally directed by user instructions similar to “Make IT Minimalist” or “Add Dark mode”.
To support current design trends and implementation patterns, AnyCoder offers the combination of web search via Tavily. When the platform is activated with an API key, the platform is on the lookout for current technologies and best practice before generating code.
Users can upload images with embedded text – similar to screenshots or handwritten notes – and anycoder extract this content using Tesseract OCR. The extracted text can then be recorded in code input requests or app content.
One-click provision for the hug of facial areas
Any code enables the immediate provision of generated apps to hug facial areas. After authentication via OautH and the granting of the essential authorizations, users can provide apps as a part of their very own name of the facial account. The deployments include:
- Mobile -friendly, responsive designs
- Brands header/footer and readme
- Living, shared URL
- Edit complete property and access
This provision function now includes support for full Python apps created with Gradio and expand the tools of the tool beyond static locations. The support for Stromlit can also be under development.
For beginners and even those with technical know -how who wish to quickly turn a brand new project, AnyCoder appears to be an awesome and convincing place to start out.
