Hi there,

It’s been a long time since our last release! We have been working very hard on Eden AI, making the platform and APIs more robust and performant. We also developed a way to release new models just a few minutes after their official launch (as we did for Claude 3.7 sonnet and GPT-4.5).

We are working on even more improvements and some very exciting new features 🤩

New Features and Enhancements

New Chat Models

We've implemented several new models in our Chat and Multimodal Chat APIs:

• OpenAI GPT-4.5 and o3-mini
• Claude 3.7 Sonnet
• Deepseek models R1 and V3
• xAI models grok-2 and grok-2-vision
• Mistral Multimodal model: pixtral-large-latest
• Qwen 2.5-72B-Instruct-Turbo
• Microsoft WizardLM-2-8x22B
• All Amazon Nova models (Pro, Lite, Micro)

New provider on Video Question Answering! 📽️

Video Question Answering is one of our latest popular API, so we wanted to add alternatives to Gemini models: we added Amazon Nova Pro model this API.

xAI grok-2 is available on Text features!

Grok-2 model performs very well in different LLM benchmarks. We wanted users to access it for text analysis tasks like: sentiment analysis, summarization, topic extraction, translation, named entity recognition, code generation, anonymization, etc.

New feature: Image Deepfake Detection!

We were already offering Image AI detection API on Eden AI but we wanted to provide something more specialized for detecting AI-generated faces. That’s why we added SightEngine’s Deepfake detection API!

Create API key and Monitor usage through API! 🔑

On Eden AI you can create multiple API keys in case you want to be able to track and monitor consumption of each users / customer or each project you have. You can now create API keys automatically through our new API endpoint, and also check consumption and credit left through our cost monitoring API.

What are we cooking for the next release? 🧑‍🍳

RAG Custom Chatbot V2!

We are currently working on a new version of our RAG feature. New intuitive UX, improve file and URL embeddings, multimodal RAG, new embedding models, more chunk management options and much more…

New LLM Chat Endpoint

We currently have two different endpoints, one for LLM supporting text input only and one for Multimodal LLMs. Those two endpoints have different models and structure, which is not always easy to use.

We are working on a new llm/chat endpoint that will support all the latest LLM models, multimodal or not —one endpoint, one document, one API and one output structure, for all models.

Plus, this endpoint will be OpenAI-compatible, which will simplify the integration of Eden AI API into third-party services.

Custom Prompt creation and versioning

We are working on a new and exclusive very powerful feature: Custom Prompt creation and versioning. It will allow you to create prompts using all the latest LLM models, save them, create different versions, choose which version to put in production, customize output format, analyze prompt performance and accuracy, and much more…

Stay tuned! 📲

We hope these updates enhance your experience with Eden AI and empower you to achieve more with our platform. Thank you for your continued support and feedback.

Do not hesitate to suggest ideas and features on our open roadmap!

Jérémy - The product guy