🚀 Download BlackBird now: build local AI agents that think, see, and reason, all offline. ✨ Try Pro for unlimited high-res images, docs, and data processing. 🎟 Use coupon code for early access: decompute.run/releases Own your intelligence. On your device.

Should we open source SDK for BlackBird’s agent building functionality?

Generate images locally with a single prompt! Bring your childhood favorite comic characters alive with your imagination with BlackBird now! Download now: decompute.run/releases #genai #image #OnDeviceAI #localAI #AIAgent

We launched BlackBird for Windows for just 4 GB VRAM. Now fine-tune Qwen-0.6B parameter model locally on your laptop (Made possible by LaserTune by Decompute AI). Download now: decompute.run/releases #ai #edgeai #onDevice #tech

BlackBird now captures all your meetings and remembers every conversation, entirely on your device. Your personal model keeps everything private and local, so you can revisit any discussion, find key insights, and ask questions anytime, privately & offline! Download here:

🚀 Introducing Decompute AI Kestrel STS 1.1B: The first low-parameter audio-reasoning model that operates in real-time entirely on-device, self-improving upon deployment. This breakthrough eliminates the traditional 3-stage voice AI pipeline (ASR → cloud LLM → TTS), removing

We believe the path to general intelligence demands models that see, hear, and act in the world around us. Text alone can’t capture the crackle in a colleague’s voice or a sarcasm in a remark, the warning beep of machinery, or the ambient clues that define context. Kestrel is our

Decompute Kestrel-SMoHE 1.0 (6B→2B active) just posted SOTA results for on-device summarization, Q&A & translation: • MMLU 0.591  • HAERAE 0.715 • PIQA 0.768  • WMT24-KR 0.717, ZH 0.618 ✈️4 GB RAM, 437 t/s Outperforms Gemma-1B, Qwen-1.7B, HyperCLOVA-1.5B & Kanana-2.1B

Confused when to use a single dense model, multiple SLMs and when to use SMoEs? A dense 6B model is like a generalist employee who “knows a bit of everything.” But when you need legal precision, you want a lawyer, not someone who also handles IT tickets and HR disputes. That’s

My way of motivating the team to find Decompute BlackBird SDK bugs. 🥳🤣🍕#DecomputeRocks 🤘🏻

Finally: AI that doesn't spy on your users. Build it with BlackBird 👇 **BlackBird SDK Launched!** Build private, on-device AI agents in minutes. No cloud, no compromises. **Why it’s game-changing:** 🔒 **100% Private**: Data never leaves the device ⚡️ **<100ms

📢 Exciting news! The BlackBird-SDK is now open source! 🎉 - Local AI inference | - Real-time streaming | - Secure file processing 👉 Contribute: github.com/Decompute-AI/B… #OpenSource #AI #MachineLearning

🚨 Calling all builders in SF! $50K up for grabs! 🚨 We’re thrilled to sponsor the A2A Agents Hackathon in SF! 💥 Build on-device AI agents with the Blackbird SDK 🧠 Multimodal, fast, secure, insanely powerful 🏆 Judged by Decompute’s own Hina Dixit 🎯 Sign up:

🚀 Introducing Kestrel VLM: our vision-language model that self-improves after deployment and leaps 30 % past DeepMinds' SigLIP-2, unlocking a new “fusion-space” frontier for enterprise vision AI. Kestrel can learn from just 5-10 examples, unlocking reinforcement learning at the

We recently shipped Kestrel VLM - a self-improving, optimized Vision-Language model for macOS, Windows, Android and iOS, available in three sizes (650M, 1B and 1.5B). Achieves ~817/1000 for OCRB, rivaling 32B-param giants, yet runs entirely on-device. Perfect for document

🚀 Exciting news! We’ve just shipped a new update for the BlackBird SDK, packed with powerful features driven by your feedback. Now, building on-device AI agents is easier and more seamless than ever — all with just a few lines of code. Here’s what’s new: ✨ Support for

1/ Kestrel VLM 1.5B — 841/1000 on OCR, 63.5% on Real-World QA and 89% on DocVQA eval!! One compact model, three hard tasks. #VLM #OCR #DocVQA #Multimodal #SLM