google-deepmind 开源项目推荐（63 个）

aduboue /img-studio

ImgStudio is a NextJS web app designed for easy deployment and user-friendly experience, streamlining access to the power of Google's GenAI model Imagen & Veo to generate powerful images & videos 🔥

skito /aipi-php

Universal API client for common AI models

zebbern /system_prompts_leaks

来自热门聊天机器人如ChatGPT、Claude和Gemini的全新系统提示词

199-biotechnologies /gemma-4-abliterated

Gemma 4 31B Abliterated — quality-preserving guardrail removal for Google's most capable open model. Apache 2.0. Runs on Apple Silicon via MLX.

jorgemunozl /psiformer_torch

self-attention neural torch ansatz for electronic structure calculations introduced in the paper “A Self-attention Ansatz for ab initio Quantum Chemistry” from Google Deep Mind.

SafeMantella /claude-code-agy-CLI-skill

Multi-agent skill for Claude Code that delegates tasks to Google Antigravity CLI — web search, image gen, codebase analysis, 40+ science databases

LikeACloud7 /awesome-frontier-ai-papers

每日追踪前沿 AI 实验室论文、模型卡、系统卡和技术报告

reis428 /system_prompts_leaks

🛠️ 发现并探索一系列系统提示和消息，有助于更好地理解和开发实践。

Anil-matcha /Veo-4-API

Python wrapper for Veo 4 API by Google DeepMind — native 4K AI video with integrated audio, character consistency & advanced camera controls.

jayeshmepani /Google-AI

This catalogue currently covers 126 major Google AI services, tools, experiments, and features across all categories.

somewherelostt /surfyy

Screen-aware desktop agent that turns voice, screenshots, docs, and current context into live Gemma 4 + Cerebras task graphs.

The-Adimension /ATRIA-EchoTrace

Fine-tune MedGemma-1.5-4B-it (via QLoRA) to output polygon coordinates for tracing the cardiac chambers on echocardiographic frames. More info at: https://discuss.ai.google.dev/t/atria-echotrace-fine-tuning-medgemma-1-5-for-polygon-based-heart-structure-contouring/172907 and https://atria.theadimension.com/

Rayhanpatel /functiongemma-hackathon

3-Tier hybrid AI router that orchestrates FunctionGemma-270M on-device and Gemini 2.5 Flash Lite in the cloud for 99% function-calling accuracy at 548ms avg latency. Built at the Cactus × Google DeepMind Hackathon.

mountsea-ai /veo-api

🎬 Google Veo 3 API & SDK — Cheapest Veo API for text-to-video & image-to-video. Veo 3 / Veo 3.1 / Veo 3 Fast supported. Python & Node.js SDK. Up to 50% OFF!

RobinaMirbahar /gemini-4-argon-cookbook

Gemini 4 Argon Cookbook — 1M Context Continuation, Deep Reasoning Thought Signatures & Interactive 3D WebGL Studio — by Robina Mirbahar

47thtechcorner /RayCodes_EmbeddingGemma2

Google's New 500MB 4-Modality AI That Runs on Phones! (EmbeddingGemma 2 Review) - On-device 740M multimodal embedding model for semantic search, codebase indexing, and cross-modal media retrieval.

crypto-code /BigGAN-Pretrained

Uses a Pretrained Network (using Google's Deepmind) to generate images.

adityapatel1010 /Physics-Informed-Graph-Neural-Network

Implemented a Physics-Informed Graph Neural Network (GNN) inspired by Google DeepMind’s research to simulate particle dynamics with high accuracy (96.2%). The model leverages PyTorch, GNNs, and transfer learning to capture complex physical interactions in particle systems using graph-based representations.

balyakin /synthid-remover

Experimental SynthID image watermark removal

akajammythakkar /gemma-on-jax

Run Google DeepMind's Gemma open models on TPU with the official JAX/Flax library. A working Colab notebook for inference and fine-tuning.