Stream

Today, we’re releasing the first version of our Vision AI framework! It's 100% open-source and has over 10 out-of-the-box plugins for @krispHQ, @ultralytics, @cartesia, @OpenAI, @GoogleAIStudio, and more! With Vision Agents, developers can build fast, low-latency vision AI applications running on Stream from day one with just a few lines of Python ⚒️

Add realtime video calling to any app in 2 minutes with Stream Skills + your favorite coding agent (Fable 5 or GPT-5.6 Sol).

Get started in just 2 steps. 1. Let a coding agent do it for iOS, Android, Flutter, and React Native Docs: getstream.io/video/sdk/ Install Stream Agent Skills from skills.sh. - Stream CLI Skill: skills.sh/getstream/agen… - iOS/SwiftUI/UIKit: skills.sh/getstream/agen… 2. Use the skills in Cursor, Codex, Antigravity, Claude Code, and more. /stream-cli, /stream-swift

Get started: - getstream.io/chat/sdk/ios/ Let a coding agent do it: Install Stream Agent Skills from skills.sh and use them in Cursor, Codex, Antigravity, Claude Code, and more. Stream CLI Skill: - skills.sh/getstream/agen… iOS/SwiftUI/UIKit: - skills.sh/getstream/agen…

Want to solve math with Math Notes in chat messaging? 👉 Stream Chat iOS/SwiftUI. It is great for day-to-day budgeting and for planning events in 1-1 and group chats.

Install Stream CLI and Swift Skills - skills.sh/getstream/agen… - skills.sh/getstream/agen… Next, prompt your coding agent to create a SwiftUI chat messaging app using the skills and the Core AI models to refine draft messages before sending. With a starter iOS/SwiftUI app, your coding agent will build a fully functioning iOS messaging app with Stream’s backend and integration of the new Apple Core AI framework. Let’s begin: Stream Chat iOS/SwiftUI: - getstream.io/chat/sdk/ios/ Core AI: - developer.apple.com/documentation/… Official Core AI Models: - github.com/apple/coreai-m… Core AI Model Zoo (LLMs converted to Core AI): - github.com/john-rocky/cor…

Improve, refine, and style draft messages with local on-device AI models before sending in chats. With Stream Chat iOS, you can easily bring Core AI models for this purpose. In this demo, we have Apple Intelligence (Foundation Models), Qwen-3.5, LFM-2.5, and Granite 4.0-H 1B, all running locally on iPhone. Want to try it out? 👇

All the models are running locally with no token cost. - UI + reusable SwiftUI chat components: getstream.io/chat/sdk/ios/ - Core AI: developer.apple.com/documentation/… - Official Core AI Models: github.com/apple/coreai-m… - Core AI Model Zoo: github.com/john-rocky/cor…

You can now build an in-app experience like ChatGPT in WhatsApp with the new Apple Core AI framework and Stream Chat iOS/SwiftUI SDK. Bring any on-device AI model you prefer. Want to try it out? 👇 What’s in this demo? - Write messages with your voice: SpeechAnalyzer API, SpeechTranscriber model, & Whisper. - Refine, summarize, fix grammar, or style messages before sending: Apple Intelligence, Foundation Models, & Qwen3.5. - Perform on-device vision tasks: Ask about images and documents with MiniCPM-v 4.6

Build it yourself: getstream.io/blog/xcode-age…

As of Xcode 26.3, you can now run agents directly inside the Xcode IDE. This makes it an amazing tool to build Swift apps, but if you're using SDKs without giving your agent access to every method, it's going to guess. We gave it our current docs as a skill instead. Now one prompt can get you a working SwiftUI chat app that actually hits the Stream API. Link to make your own Stream app in the thread 🔗

SwiftUI: You can chain free and open-source STT, TTS, and LLMs to build voice/vision apps with Core AI. Here is a demo I was experimenting with: - Whisper: Speech-to-text - Kokoro-82M: Text-to-speech - Qwen3-VL 2B: Vision I converted Pocket TTS and VibeVoice to Core AI models and tried with Gemma 4. All is working great.

Get started by scaffolding your app with the Stream Agent Skills and enabling the voice messaging feature with a single line of code. skills.sh/?q=getstream composerConfig: ComposerConfig(isVoiceRecordingEnabled: true)

Voice messaging is available in our iOS/SwiftUI, Android/Compose, Flutter, and React Native SDKs. - getstream.io/chat/sdk/ios/, - getstream.io/chat/docs/sdk/… - getstream.io/chat/sdk/andro… - getstream.io/chat/sdk/flutt… - getstream.io/chat/sdk/react…

Want users to record, playback, send, and receive voice messages in your app? 👇

To get started quickly: 1. Install these skills & and use your agent to build an iOS/SwiftUI app with Core AI and Stream Chat npx skills add github.com/getstream/agen… --skill stream-cli skills.sh/getstream/agen… getstream.io/chat/sdk/ios/ 2. Find Core AI models to run Official Core AI models github.com/apple/coreai-m… Core AI Model Zoo github.com/john-rocky/cor… 3. Install for example, the MiniCPM-V-4.6 vision model in your SwiftUI app huggingface.co/mlboydaisuke/M… github.com/john-rocky/cor… You are now up and running with an on-device vision-language model locally on your iPhone / Apple Silicon Mac. No API calls, no token cost, no model subscription.

The best and natural way to interact, run, and test on-device AI models locally on iPhone/iPad/Apple Silicon Mac? Core AI + Stream Chat SwiftUI. Want to try it out? 👇

Want to try it out in Stream Chat iOS/SwiftUI? Install these agent skills. npx skills add github.com/getstream/agen… --skill stream-cli skills.sh/getstream/agen… Grab the Core AI models. huggingface.co/mlboydaisuke/M… github.com/john-rocky/cor… Build an iOS/SwiftUI Stream Chat app with the above agent skills and Core AI integration using your favorite coding agent, such as Fable, and a specified Core AI model. getstream.io/chat/sdk/ios/ Official Core AI models github.com/apple/coreai-m… Core AI Model Zoo github.com/john-rocky/cor…

If you get stuck the moment you want an on-device model to see something, a PDF, a chart, a document, an image, or a video, try MiniCPM-V-4.6 in Apple’s new Core AI framework. Surprisingly, we found this to be the best on-device vision-language model for iPhone / Apple Silicon. Run it fully locally on your iPhone or Mac with Core AI and ready-made Stream Chat SwiftUI components. Pick a photo, doc, or video and ask about it. MiniCPM-V-4.6 will quickly stream an answer.

@pattssun @nyknicks Nice shot!

The @nyknicks championship inspired me to build an AI basketball coach to improve my 1v1 game Built with: - Robloflow RF-DETR to detect players/ball/rim - MediaPipe to capture body angles - OpenCV to analyze and annotate the clip