Deep dives, tutorials, and analysis for AI-powered creators.
Microsoft launched three in-house AI models on April 2, reducing its dependence on OpenAI. MAI-Image-2, MAI-Voice-1, and MAI-Transcribe-1 cover image generation, text-to-speech, and speech recognition.
Alibaba releases Qwen3.6-Plus with 1M-token context, autonomous repository-level coding, and the ability to convert wireframes and screenshots directly into working frontend code.
GitHub launches /fleet in Copilot CLI, dispatching multiple AI agents to work on different tasks simultaneously instead of processing requests sequentially.
Liquid AI releases LFM2.5-350M, a 350M parameter open-weight model that runs AI agents on mobile GPUs in just 81MB with 32k context window.
Z.ai launches GLM-5V-Turbo, a vision-coding model that scored 94.8 on Design2Code versus 77.3 for Claude Opus 4.6. Converts mockups directly to frontend code.
OmniVoice is a zero-shot text-to-speech model supporting 600+ languages with voice cloning, an Apache 2.0 license, and inference 40x faster than real-time.
Alibaba released Wan2.7-Image, a unified AI model that handles both image generation and editing in one system, with fine-grained personalization, color control, and a Pro variant capable of 4K output.
Arcee AI releases Trinity-Large-Thinking under Apache 2.0, ranking #2 on PinchBench behind Claude Opus 4.6 at just $0.90 per million output tokens.
PrismML emerged from stealth and open-sourced Bonsai 8B, a true 1-bit large language model that fits 8.2 billion parameters into just 1.15 GB of memory, roughly 14x more compressed than comparable 16-bit models.
Technology Innovation Institute releases Falcon Perception, a compact 0.6B parameter vision model that beats Meta SAM 3 across six key benchmarks while running on consumer hardware.
Google launches Veo 3.1 Lite, a cost-effective video generation model available now via the Gemini API and Google AI Studio.
H Company released Holo3, a vision-language model optimized for GUI agents that sets new state-of-the-art on the OSWorld-Verified benchmark, beating proprietary models with only 3B active parameters.
Anthropic accidentally exposed the full source code of Claude Code on March 31, leaking roughly 1,900 files and over 500,000 lines of TypeScript through an npm packaging error.
Adobe Illustrator Turntable is now generally available, automatically generating up to 74 editable 3D views from a single 2D vector.
Alibaba releases Qwen3.5-Omni, a multimodal AI model that processes text, images, audio, and video while generating real-time speech output in 36 languages.
Skywork AI releases Matrix-Game-3.0, an open-source model that generates real-time interactive video at 720p and 40fps.
Apple researchers published LGTM on March 28, 2026, a method that enables 4K-resolution 3D scene rendering using Gaussian Splatting without requiring per-scene optimization.
Anthropic confirmed that Claude paid subscriptions have more than doubled in 2026, with credit card data showing record subscriber growth in January and February.
NVIDIA has open-sourced Kimodo, the largest controllable motion diffusion model ever trained. Trained on over 700 hours of professional motion capture data, it generates 3D human and robot motions from text prompts.
Meta releases SAM 3.1 with object multiplexing that tracks 16 objects simultaneously at 32fps on a single H100 GPU.