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 ⚒️

🔗 Activity Feeds changelog: getstream.io/activity-feeds…

Stream Q1 product updates 🔔 Major improvements across all our APIs — starting with Activity Feeds. Ranked feeds, the engine behind "For You" style timelines, got: - Faster, more reliable pagination - No default cutoff window (more flexible ranking) - Simpler custom ranking (recency now default) - Edge case fixes for autocomplete Full details in thread.

@visionagents_ai 🔗 Build a local AI agent in < 5 minutes: getstream.io/blog/qwen3-loc…

Build a local vision + voice agent with Qwen 3.5 Small. Runs entirely on your laptop using Ollama + Python. No cloud LLM calls. Stack: - Qwen 3.5 Small - Stream WebRTC - Deepgram + ElevenLabs - @visionagents_ai Thread with code ↓

@visionagents_ai @bvsbharat @dibbaa_code @GretchenBoria Explore all of the builds + code + tutorials here ↓ getstream.io/blog/gemini-3-… We can't wait to see what you'll create!

@visionagents_ai @bvsbharat @dibbaa_code 👀 Scrym Vision by @GretchenBoria & Jorge Jimenez A tactical dashboard that detects formations and maps player movement live, alongside win probability and dynamic game visuals. 🎥 Watch the demo: youtube.com/watch?v=1YZDAI…

At the Gemini 3 SuperHack in San Francisco, we watched developers build some seriously creative AI projects with @visionagents_ai. Seeing people experiment with vision AI (and ship in a single day) was easily the best part. 👇 Check out 3 of our favorite builds + demos in the thread.

This is @GoogleDeepMind Gemini 3.1 Flash-Lite responding in real time in a Vision Agents app. It's able to handle a lot of different video understanding questions much more quickly than the previous gen... and this is on release day, when everyone's hitting the API! 😆

@Kimi_Moonshot @visionagents_ai 🔗 Full tutorial: getstream.io/blog/kimi-k2-a…

We built a real-time video + vision + voice agent using @Kimi_Moonshot K2.5 and @visionagents_ai. The agent analyzes live camera feed, answers questions on what it sees, and helps with coding tasks. Build it yourself in under 5 minutes ↓

Expo SDK 55 is OUT! It comes with a ton of improvements and updates. So I made a Full Stack tutorial where we build a video calling app powered by @expo @getstream_io @sentry @clerk Here's a 5-minute demo👇 Full project link below.

There's still time to join us :)

Decided to do this week's changelog as a video, key things mentioned are in the 🧵:

Day 5 of the @VisionAgents Vision Possible hackathon by @WeMakeDevs . 48 hours left. Here's what RoundZero looks like right now 👇 🧠 What is RoundZero? An AI mock interviewer that watches your face, listens to your answers, and interrupts you mid-sentence when you go off track.

Stream Chat architecture + benchmarks ↓ getstream.io/chat/docs/reac…

We benchmarked chat at extreme scale. Results: - 5M concurrent users in one channel - Sub-40ms latency - Stable performance throughout the test - No hard caps on scale Infra behind it: - Edge auth + rate limiting at the edge - Cuckoo filters + optimized WS broadcast - writev + zero-copy message delivery - Redis cluster with atomic Lua ops Full breakdown in thread. ⬇️

Follow the full tutorial: Build an Instagram-Style For-You Feed in React Native 🔗 getstream.io/blog/instagram…

Posting content uses a dedicated composer: • Text input + media attachments • Activities written to the `user` feed • Fan-out handled automatically to timelines and discovery feeds This keeps write logic simple and read paths fast.

Want to build an Instagram-style "For You" feed? Follow this tutorial to structure feeds, rank content by popularity, and wire up real-time updates using Expo + Stream's React Native Feeds API.