Quanta AI Labs

18 / Apr 28 – May 4, 2025 · 10 RELEASES

AI News Week 18, 2025

Week 18 2025 AI releases.

Vision & Image (1)

  • Seed1.5-VL

    ByteDance · Visual Understanding

    A vision-language model supporting multimodal understanding and generation across images and text. Extends the Seed model family with improved visual reasoning and semantic alignment capabilities.

Audio & Speech (1)

  • ARC Text-to-Audio Generation

    UC San Diego, Stability AI, Arm · Text-to-Speech

    Introduces Adversarial Relativistic-Contrastive (ARC) post-training, a novel acceleration algorithm for diffusion/flow models that generates 12 seconds of 44.1kHz stereo audio in ~75ms on H100 GPUs and ~7 seconds on mobile edge devices. Combines relativistic adversarial formulation with contrastive discriminator objective to improve prompt adherence without distillation.

3D & Spatial (1)

  • Step1X-3D

    StepFun AI / LightIllusions Team · 3D Generation

    Open-source framework for high-fidelity 3D asset generation combining a hybrid VAE-DiT geometry generator with SD-XL-based texture synthesis, curated from 2M high-quality assets with support for controllable generation and 2D control technique transfer.

Agents & Automation (1)

  • AlphaEvolve

    Google DeepMind · Coding Agents

    A Gemini-powered coding agent that automatically designs and optimizes advanced algorithms through evolutionary search and iterative refinement. Combines large language models with algorithmic optimization techniques to discover novel computational solutions.

Developer Tools (6)

  • llama.cpp Multimodal

    ggml-org · Inference Engine

    llama.cpp multimodal documentation enabling vision-language capabilities for local LLM inference with support for image inputs alongside text processing.

  • ComfyUI-LBMWrapper

    kijai · AI Infrastructure

    A wrapper integration for Latent Basis Models (LBM) within ComfyUI, enabling latent space manipulation and image generation through a node-based interface. Provides modular access to LBM capabilities for generative workflows.

  • SmolVLM Realtime Webcam

    ngxson · Inference Engine

    Real-time vision-language model implementation for webcam input using SmolVLM, enabling live multimodal inference on video streams with minimal computational overhead.

  • Comfy Runner

    piyushK52 · AI Infrastructure

    A runner/executor for ComfyUI workflows that enables programmatic execution and automation of node-based AI image generation pipelines. Provides API and scripting capabilities to run ComfyUI graphs without the UI.

  • n8n

    n8n · AI Infrastructure

    n8n is a workflow automation platform that enables users to connect and automate applications through a visual node-based interface. It provides infrastructure for building complex automation workflows without requiring extensive coding.

  • RunPod Worker ComfyUI

    RunPod · AI Infrastructure

    A worker implementation for running Comfy UI (a node-based generative AI interface) on RunPod's serverless GPU infrastructure. Enables distributed deployment of ComfyUI workflows with automatic scaling.

Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.