← Voltar

πŸ“° NotΓ­cias de 2026-04-08

ia-updates

Wan AI: New Creative Platform for Text-to-Video

Wan AI is an emerging creative platform offering text-to-image, image-to-video, and text-to-video generation. The platform aims to lower barriers to AI-assisted video creation, with Wan 2.6 Flash pric

πŸ”— Fonte principal
ia-updates

Stanford HAI Releases 2026 AI Index

Stanford's Human-Centered AI Institute published its comprehensive AI Index report, providing data-driven insights into AI trends, investments, and societal impact. The index tracks progress across re

πŸ”— Fonte principal
ia-updates

Sora 2 App and API Shutting Down

OpenAI announced that the Sora 2 app and API will be discontinued in April 2026, marking the end of the consumer-facing Sora product line. Source: Wikipedia - https://en.wikipedia.org/wiki/Sora_(text-

πŸ”— Fonte principal
ia-updates

Seedance 2.0: Intent-Driven AI Video Generation

Higgsfield's Seedance 2.0 platform enables cinematic video generation from text prompts and images, featuring multi-camera storytelling, frame-level precision, and synchronized audio. It has attracted

πŸ”— Fonte principal
ia-updates

OpenClaw 2026.4.5 Released with AI Video Generation

The latest OpenClaw release (April 5, 2026) introduces AI video generation capabilities and a "dreaming memory" feature, enabling more autonomous agent behaviors. This marks a significant step toward

πŸ”— Fonte principal
ia-updates

OpenAI GPT-5.4 Thinking Achieves 75% on OSWorld-Verified, Surpasses Human-Level

The "Thinking" variant of GPT-5.4, leveraging test-time compute, has achieved 75.0% on the OSWorld-Verified desktop task benchmark, officially surpassing human-level performance. It also solved all 30

πŸ”— Fonte principal
ia-updates

MIT Technology Review: AI Benchmarks Are Broken

Critics argue that current AI benchmarks are flawed, with models achieving impressive scores that don't reflect real-world capability. The article calls for new evaluation methods that better measure

πŸ”— Fonte principal
ia-updates

Microsoft Launches MAI Model Suite

Microsoft released three foundational AI models: MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2, deepening its AI strategy with proprietary capabilities for transcription, voice, and image tasks. Sour

πŸ”— Fonte principal
ia-updates

Gemma 4 and Kimi K2.5 Highlight Open-Source Advances

Google's Gemma 4 (Apache 2.0, released April 2026) and Moonshot's Kimi K2.5 are leading open-weight models. Gemma 4 targets on-device AI across Android and smartphones, while Kimi K2.5 pushes frontier

πŸ”— Fonte principal
ia-updates

AutoGPT: State of AI Video Generation April 2026

Updated ranking of video generation models by ELO scores across multiple arenas (Artificial Analysis, arena.ai, designarena.ai). Highlights include Kling 3.0 and Wan 2.6 Flash as top contenders for im

πŸ”— Fonte principal