2026年最佳开源AI工具

Image

ComfyUI

ComfyUI is a free, open-source node-based UI for running Stable Diffusion, FLUX, and video generation models locally. It gives power users complete control over every step of the generation pipeline, from model loading to sampling to upscaling.

AI Agents & Automation

Hermes

Hermes Agent is an open-source, self-hostable autonomous agent from Nous Research. It learns from every task, works with any LLM provider, and hit 100,000 GitHub stars in 53 days. The sycophancy bug and v0.x instability are real, here is what that actually means.

AI Models & LLMs

DeepSeek

DeepSeek offers frontier-level reasoning and coding performance at API costs 10–30x cheaper than GPT-4o. The tradeoff is real: political censorship on the hosted product, PRC data residency, and a documented database exposure. Open weights mean local deployment is possible, at a price.

AI Agents & Automation

n8n

n8n is a fair-code workflow automation platform that lets developers and technical teams build complex automations and AI agents, either self-hosted for free or on n8n Cloud. With 500+ native integrations and MCP support, it is the leading open-source alternative to Zapier and Make.

Image

FLUX

FLUX is an open-weight text-to-image model from Black Forest Labs, the team behind Stable Diffusion. It launched August 2024 with three tiers, beat Stable Diffusion 3 and Midjourney v6 on benchmark leaderboards, and anchors one of the largest LoRA ecosystems in AI image generation.

AI Models & LLMs

Gemma

Gemma is Google DeepMind's family of open-weight language models, downloadable and self-hostable for free. Gemma 4 (April 2026) delivers frontier-level reasoning and multimodal capabilities under an Apache 2.0 license, from 2B edge models to a 31B dense flagship.

Image

Stable Diffusion

Stable Diffusion is the open-weight image generation standard from Stability AI. Self-host for free, fine-tune on custom datasets, and use ControlNet for precise composition control. The largest open-source image model ecosystem available.

AI Agents & Automation

Skywork

Skywork is an AI workspace platform from Singapore-based Kunlun Tech that uses specialized Super Agents to produce cited documents, presentations, spreadsheets, podcasts, and websites from a single prompt. Freemium with a $19.99/month Pro plan.

Code

Onlook

Onlook is an open-source visual editor for React apps, backed by Y Combinator (W25), that lets designers point-and-click edit live Next.js and Tailwind projects while changes write directly back to the codebase with no handoff required.

AI Agents & Automation

Cal.com

Cal.com is an open-source scheduling platform built by Peer Richelsen and Bailey Pumfleet. It offers unlimited event types on its free tier, developer-grade routing and API access, and Cal.ai voice agents for automated booking calls.

AI Agents & Automation

Botpress

Botpress is an AI-native chatbot and agent builder for developers. Build complex conversational workflows with visual tools, Autonomous Nodes, LLM-powered knowledge bases, and deploy across WhatsApp, Slack, websites, and more.

Text

Ghost

Ghost is an open-source publishing platform built for independent creators, journalists, and publications. It combines blogging, email newsletters, and paid memberships in one tool, with 0% revenue fees and full self-hosting support since 2013.

AI Models & LLMs

Jan

Jan is a free, open-source desktop app by Menlo Research that runs large language models entirely on your own hardware. No cloud, no subscriptions, no data leaving your machine. Supports Llama, Qwen, Mistral, Gemma, and dozens more GGUF models on Windows, macOS, and Linux.

Code

Axolotl

Axolotl is a free, open-source LLM fine-tuning framework that lets ML teams run reproducible LoRA, QLoRA, and GRPO training runs on 100+ model architectures using a single YAML config file across multi-GPU setups.

AI Models & LLMs

Cherry Studio

Cherry Studio is a free, open-source desktop client that unifies 300+ LLM providers including OpenAI, Claude, Gemini, and local Ollama models in a single interface. It adds built-in RAG knowledge bases, MCP agent support, and side-by-side model comparison on Windows, macOS, and Linux.

Data

RAGFlow

RAGFlow is InfiniFlow's open-source enterprise RAG engine, built around a neural document parser (DeepDoc) that handles complex PDFs, tables, and mixed-layout documents. Self-hostable, Apache 2.0, with 79k+ GitHub stars.

Code

DeepSpeed

DeepSpeed is Microsoft's open-source deep learning optimization library for training and running large language models across multiple GPUs. It powers BLOOM 176B, Megatron-Turing 530B, and countless fine-tuning pipelines via its ZeRO memory optimizer.

Code

BentoML

BentoML is an open-source Python framework for packaging and deploying machine learning models as production APIs. It supports any ML framework, includes OpenLLM for self-hosted LLM serving, and offers BentoCloud for managed inference with autoscaling and BYOC.

Data

MLflow

MLflow is the open-source AI engineering platform used by thousands of enterprises to track experiments, manage model registries, and deploy classical ML and GenAI applications. Free under Apache 2.0, with 30 million monthly downloads.

AI Models & LLMs

MLX

MLX is Apple ML Research's open-source array framework for running and fine-tuning machine learning models on Apple Silicon. Free, MIT-licensed, and designed to extract maximum performance from M1 through M5 chips using unified memory.

Code

Unsloth

Unsloth is an open-source fine-tuning library by Daniel and Michael Han that delivers 2x faster LLM training and up to 70% less VRAM through custom Triton CUDA kernels. Supports 500+ models including Llama, Qwen, and DeepSeek. Free to use, with a paid Pro tier for multi-GPU training.

AI Models & LLMs

llama.cpp

llama.cpp is the open-source C++ inference engine that powers Ollama, LM Studio, and most local LLM tooling. Created by Georgi Gerganov in 2023, it runs 50+ model architectures on any hardware, including consumer laptops and Raspberry Pis.

Audio

LiveKit

LiveKit is the open-source WebRTC infrastructure and AI agent framework that powers ChatGPT's Voice Mode. Founded in 2021, it gives developers full control over real-time voice and video pipelines, from self-hosted media servers to production-ready agent orchestration.

Audio

Pipecat

Pipecat is an open-source Python framework by Daily for building real-time voice and multimodal AI agents. It chains STT, LLM, and TTS services into programmable pipelines with full developer control, supporting 100+ integrations including Deepgram, OpenAI, ElevenLabs, and LiveKit.

Code

Quivr

Quivr is an open-source Python library for adding retrieval-augmented generation to any application. Apache 2.0, self-hostable, supports OpenAI, Anthropic, Mistral, and Groq. Built by a YC W24 team, with 39,000+ GitHub stars since its 2023 launch.

Code

Haystack

Haystack is deepset's open-source Python framework for building production-ready RAG pipelines and AI agents. Apache 2.0 licensed, it gives developers explicit control over retrieval, context routing, and generation through composable, YAML-serializable pipeline graphs.

Code

Langflow

Langflow is an open-source visual builder for LLM pipelines, RAG applications, and multi-agent workflows. Created in 2023 and acquired by DataStax in 2024, it lets developers assemble complex AI flows on a canvas and export them as production APIs, no boilerplate required.

Code

Dify

Dify is an Apache 2.0 open-source platform for building production-ready LLM applications. Visual workflow canvas, RAG pipelines, agent builder, and 100+ model integrations. Free self-hosted or $59/mo on cloud.

AI Models & LLMs

Zhipu GLM

Zhipu GLM is an open-weight large language model family from Beijing-based Z.ai (Zhipu AI), built on a mixture-of-experts architecture. MIT licensed and self-hostable, the GLM-4.5 and GLM-4.6 generations offer 200K-token context, free API flash tiers, and benchmark results competitive with Claude and DeepSeek.

AI Agents & Automation

CAMEL-AI

CAMEL-AI is the open-source multi-agent framework behind the OWL agent, which ranked #1 among open-source systems on the GAIA benchmark. Built on NeurIPS-published research from KAUST, it supports role-playing agents, 1M-agent society simulation, and autonomous task pipelines.

AI Agents & Automation

Trigger.dev

Trigger.dev is an open-source TypeScript platform for building background jobs and AI agent workflows without serverless timeout limits. Apache 2.0, with a managed cloud and self-hosting option.

AI Agents & Automation

Activepieces

Activepieces is an MIT-licensed workflow automation platform with a visual step builder, 280+ integrations, and a free self-hosted community edition. Paid cloud plans start at $25/month with unlimited task execution across all tiers.

AI Models & LLMs

Pinokio

Pinokio is a free desktop app by cocktailpeanut that turns installing open-source AI tools into a one-click operation. It manages all Python environments, dependencies, and GPU setup automatically, making ComfyUI, Whisper, Bark, and hundreds of other AI apps accessible to anyone without terminal experience.

Audio

Sesame AI

Sesame AI is the startup behind Maya and Miles, two voice companions powered by the Conversational Speech Model (CSM). The 1B-parameter open weights released March 2025 under Apache 2.0 let developers self-host a speech model that reproduces natural conversation rhythm, pauses, and disfluencies.

Code

Stagehand

Stagehand is an MIT-licensed TypeScript SDK from Browserbase that adds natural-language browser control (act, extract, observe) on top of CDP, letting developers build browser agents that survive UI changes without selector maintenance.

Code

Vercel AI SDK

Vercel AI SDK is an open-source TypeScript toolkit for building AI-powered applications and agents. It provides a unified API across 20+ model providers, streaming UI hooks for React and Vue, structured output with Zod, and a full agent abstraction layer.

Code

SWE-Agent

SWE-Agent is an open-source autonomous software engineering agent from Princeton University that uses a custom Agent-Computer Interface to resolve GitHub issues, fix bugs, and tackle cybersecurity challenges. Free, MIT-licensed, model-agnostic.

Code

OpenHands

OpenHands is an open-source autonomous AI coding agent by All Hands AI. It plans, writes, executes, and debugs code inside a sandboxed environment without requiring manual approval at every step. Free to self-host; cloud tier available.

Code

Genkit

Genkit is Google's open-source AI framework for TypeScript, Go, and Python developers. It provides typed flows, RAG, tool calling, and multi-agent support with a standout local Developer UI for iterating prompts and debugging traces. Free under Apache 2.0.

Code

OpenAI Agents SDK

OpenAI's official open-source framework for multi-agent workflows. Launched March 2025 as the production successor to Swarm. Provides handoffs, guardrails, tracing, and voice agent support in Python and TypeScript.

Code

OpenAI Codex CLI

OpenAI Codex CLI is an open-source terminal agent that reads your codebase, edits files, and runs shell commands with OS-level sandboxing. Released April 2025, rewritten in Rust, free under Apache 2.0.

Code

Agno

Agno is an open-source Python framework for building multi-agent AI systems, rebranded from Phidata in January 2025. Model-agnostic, Apache 2.0, with built-in memory, knowledge bases, and a production-ready AgentOS runtime for deploying agents as services.

Audio

GPT-SoVITS

GPT-SoVITS is a free, MIT-licensed voice cloning and TTS system that clones a voice from just 1 minute of audio. Developed by RVC-Boss, it excels at Chinese, Japanese, and anime character voice reproduction. Popular with VTubers, fan dubbers, and indie audiobook creators worldwide.

Code

Letta

Letta is the open-source agent framework from UC Berkeley's MemGPT research team. It gives AI agents persistent, self-editable memory using an OS-inspired three-tier architecture. Apache 2.0 with a cloud-hosted API option.

Code

Roo Code

Roo Code is a free, open-source VS Code extension that runs role-based AI agents (Architect, Code, Debug, Orchestrator) directly in your editor. Apache 2.0, bring-your-own-API-key, no subscription needed. Reached 3 million installs before archiving in May 2026.

Code

smolagents

smolagents is Hugging Face's open-source Python library for building AI agents that write code as actions rather than JSON tool calls. Apache 2.0, ~1,000 lines of core code, supports 100-plus LLMs including local models via Ollama and Transformers.

Code

Zed AI

Zed is a GPU-accelerated, open-source code editor built in Rust by the creators of Atom. It pairs 120fps performance with an open AI agent protocol, letting developers bring their own models or agents rather than locking into a single provider.

Code

Mastra

Mastra is an open-source TypeScript framework for building production AI agents, from the team behind Gatsby. It ships agents, durable workflows, RAG, built-in memory with SOTA LongMemEval scores, and a visual debugging studio.

Code

PydanticAI

PydanticAI is an open-source Python framework for building production-grade AI agents. Created by Samuel Colvin (creator of Pydantic), it brings FastAPI-style type safety, structured outputs, and dependency injection to agent development. Free, MIT-licensed, 16k+ GitHub stars.

Code

DSPy

DSPy is a Stanford NLP framework that treats prompt engineering as a compiler problem: you write typed signatures and modules, define a metric, and an optimizer searches for the best prompts automatically. Free, Apache 2.0, 5M+ monthly PyPI downloads.

Audio

OpenVoice

OpenVoice is an MIT-licensed instant voice cloning model from MIT and MyShell AI. It clones a voice from a short reference clip, supports English, Spanish, French, Chinese, Japanese, and Korean, and is free to use commercially with no API fees.

Code

LiteLLM

LiteLLM is an open-source AI gateway by BerriAI (YC W23) that translates 100+ LLM provider APIs into a single OpenAI-compatible endpoint. Self-host the proxy server to route across OpenAI, Anthropic, Bedrock, and more with spend tracking, rate limiting, and fallback routing built in.

Audio

F5-TTS

F5-TTS is an open-source text-to-speech model that clones any voice from a 10-15 second audio sample using flow matching. Developed by researchers at Shanghai Jiao Tong University and Cambridge, it runs entirely on local hardware with no API costs.

Video

WAN (Wan-Video / Wan2GP)

WAN is Alibaba's open-weight video generation series (2.1 through 2.7), offering text-to-video, image-to-video, and editing capabilities under Apache 2.0. It topped the VBench leaderboard at launch in February 2025 and became the default choice for self-hosted AI video work.

AI Models & LLMs

vLLM

vLLM is an open-source LLM inference library from UC Berkeley that delivers high-throughput, memory-efficient serving for hundreds of open models. Free under Apache 2.0, with an OpenAI-compatible API and support for multi-GPU deployments.

Code

Firecrawl

Firecrawl is a web scraping API by Mendable AI that converts websites into LLM-ready Markdown and structured JSON. Built for RAG pipelines, AI agents, and developer workflows. 113,000 GitHub stars, used by 80,000+ companies.

AI Agents & Automation

E2B

E2B is open-source cloud infrastructure that gives AI agents a secure, isolated sandbox to execute code. Used by 88% of Fortune 100 companies, it runs Python, JavaScript, and any Linux-compatible language inside Firecracker microVMs with sub-200ms startup.

AI Agents & Automation

Browserbase

Browserbase provides managed cloud browser infrastructure for AI agents, pairing raw headless Chromium sessions with Stagehand, an open-source SDK that replaces brittle CSS selectors with natural language act(), extract(), and observe() primitives.

Data

pgvector

pgvector adds vector similarity search to any PostgreSQL database. Store embeddings as native columns, query with cosine or L2 distance using plain SQL. No extra service required. Used in production by teams on Supabase, AWS RDS, GCP Cloud SQL, and Neon.

Data

Milvus

Milvus is an open-source, cloud-native vector database built for billion-scale similarity search. Backed by Zilliz and used in production by NVIDIA, Salesforce, and eBay. Free to self-host. Zilliz Cloud managed service starts free with paid tiers from $99/month.

Data

PostHog

PostHog is the open-source, all-in-one analytics platform built for engineers. It replaces Mixpanel, Hotjar, LaunchDarkly, and more in a single stack, with a free tier generous enough that over 90% of teams pay nothing.

Data

Helicone

Helicone is an open-source LLM observability platform and AI gateway from YC W23. One URL swap gives teams cost tracking, request logging, semantic caching, and prompt management across OpenAI, Anthropic, and 100+ models. Used by 16,000 organizations before its March 2026 acquisition by Mintlify.

Data

Langfuse

Langfuse is an open-source LLM observability and engineering platform used by AI teams to trace, debug, evaluate, and manage prompts in production. MIT-licensed, self-hostable, and trusted by over 63 Fortune 500 companies including Khan Academy.

Video

HunyuanVideo

HunyuanVideo is Tencent's open-source video generation model, launched December 2024 with 13 billion parameters. It rivals closed-source competitors like Runway Gen-3 in visual quality, runs locally via ComfyUI, and has spawned over 500 community LoRAs on CivitAI. ---

Video

CogVideoX

CogVideoX is Zhipu AI's open-source video generation model series, producing text-to-video and image-to-video clips locally or via API. Runs on consumer GPUs from 5GB VRAM with full diffusers support, LoRA fine-tuning, and ComfyUI integration. Completely free to download and self-host.

Code

Bolt.new

Bolt.new is StackBlitz's browser-native AI app builder that turns text prompts into running full-stack web applications in minutes, using WebContainers to execute Node.js entirely inside your browser tab without any local installation.

AI Models & LLMs

AI21 Jamba

AI21 Jamba is a hybrid Mamba-Transformer model family built for long-context enterprise tasks. Open-weight models run free on your own hardware; the API starts at $0.20 per million tokens with a genuine 256K context window.

Data

Chroma

Chroma is the open-source embedding database developers reach for first when building RAG pipelines. Apache 2.0, installable in one line, and beloved for its minimal four-function API. Production scale is a known constraint worth understanding before you commit.

Data

Qdrant

Qdrant is an open-source vector search engine built in Rust by a Berlin-based team. It powers production RAG, semantic search, and recommendation systems for enterprises including Canva, HubSpot, and TripAdvisor, with Apache 2.0 licensing and flexible deployment from laptop to edge device.

Data

Weaviate

Weaviate is an open-source vector database that stores objects and embeddings together, enabling hybrid search, RAG pipelines, and AI agent workflows. Available self-hosted under BSD-3 license or as a managed service via Weaviate Cloud on AWS, GCP, and Azure.

Image

AUTOMATIC1111 Stable Diffusion WebUI

AUTOMATIC1111 is the original open-source graphical interface for running Stable Diffusion locally. Free, self-hosted, and AGPL-3.0 licensed, it made local AI image generation accessible in 2022. In 2026, it remains the best-documented starting point, though Forge and ComfyUI have since eclipsed it for serious work.

AI Models & LLMs

LocalAI

LocalAI is a free, open-source AI engine by Ettore Di Giacinto that runs any model locally with a drop-in OpenAI-compatible API. No cloud, no GPU required, and no data leaves your hardware.

AI Agents & Automation

AnythingLLM

AnythingLLM by Mintplex Labs is a free, open-source application for building private AI assistants that chat with your documents, run local models, and execute multi-step agent workflows, all on your own hardware with no cloud dependency.

AI Models & LLMs

Open WebUI

Open WebUI is a free, self-hosted interface for running local LLMs. It connects to Ollama and any OpenAI-compatible backend, adds a polished chat UI, document RAG, voice input, multi-user access controls, and enterprise SSO, all without data leaving your server.

AI Agents & Automation

Flowise

Flowise is an open-source, drag-and-drop builder for LLM workflows and AI agents. Self-host it free under Apache 2.0 or use the managed cloud. 52,000+ GitHub stars, acquired by Workday in August 2025.

AI Agents & Automation

Open Interpreter

Open Interpreter is a free, open-source CLI that gives large language models direct access to your computer's code execution environment. Describe a task in plain language, review the generated code, and watch it run locally across Python, JavaScript, and shell.

Code

Cline

Cline is a free, open-source VS Code extension that turns any LLM into an autonomous coding agent. It can create and edit files, run terminal commands, and browse the web, but asks for explicit approval before every change. You bring your own API key.

Code

Continue.dev

Continue is a free, open-source AI coding assistant that plugs into VS Code and JetBrains. Founded by YC W23 alums Ty Dunn and Nate Sesti, it lets developers bring any LLM including local models via Ollama, making it the most flexible alternative to Cursor or GitHub Copilot.

Code

Aider

Aider is an open-source, terminal-native AI pair programmer that wires any LLM directly into your local git repository. Every edit is auto-committed with a descriptive message. There is no subscription, no vendor lock-in, and no IDE required.

AI Agents & Automation

LangGraph

LangGraph is an open-source Python framework by LangChain for building production-grade AI agents as stateful directed graphs. It provides durable execution, built-in checkpointing, and human-in-the-loop controls used in production by Klarna, LinkedIn, and Uber.

AI Agents & Automation

AutoGen

AutoGen is Microsoft Research's open-source Python framework for building multi-agent AI applications. It lets developers define autonomous agents that collaborate through message passing, supporting both conversational workflows and complex agentic pipelines. Free, MIT-licensed, self-hostable.

AI Agents & Automation

CrewAI

CrewAI is an open-source Python framework for building multi-agent AI systems using role-based crews. Founded by João Moura in 2023, it powers billions of agent executions monthly and offers both self-hosted and enterprise cloud deployment options.

AI Models & LLMs

Hugging Face

Hugging Face is the world's largest open-source AI platform, hosting 2 million models, 500,000 datasets, and 1 million Spaces apps. It is the foundational hub where researchers and developers discover, fine-tune, and deploy models across every AI domain.

Audio

Coqui TTS

Coqui TTS is an open-source text-to-speech framework built around XTTS v2, a voice cloning model that clones any voice from 6 seconds of audio across 17 languages. Free to use locally for non-commercial projects; the company behind it shut down in January 2024.

Audio

AudioCraft

AudioCraft is Meta's open-source audio AI framework, bundling MusicGen for text-to-instrumental music, AudioGen for sound effects, and EnCodec for neural audio compression. Free to self-host, MIT-licensed code, ideal for researchers and developers building custom audio models.

Audio

Whisper

OpenAI Whisper is a free, MIT-licensed speech recognition model that transcribes audio in 99 languages. Used by developers for podcast pipelines, meeting notes, and video subtitles. Self-hosted at zero cost; also available via OpenAI's API at $0.006 per minute.

AI Models & LLMs

OLMo

OLMo is AI2's fully open language model family, releasing not just weights but the complete Dolma training dataset, all training code, and intermediate checkpoints under Apache 2.0. The research community's go-to for reproducible, auditable LLM science.

AI Models & LLMs

Falcon

Falcon is the UAE government-backed open-weight LLM family from the Technology Innovation Institute, spanning dense transformers, Mamba state-space models, and hybrid architectures. It is the leading open-source option for Arabic NLP, free to download and self-host.

AI Models & LLMs

Phi

Microsoft's Phi family proves data quality beats scale. The Phi-4 generation (14B, MIT license) punches above its weight on math and reasoning, and runs on consumer hardware. Free to download, with paid hosted inference on Azure AI Foundry.

AI Models & LLMs

Replicate

Replicate is a cloud API platform that lets developers run over 50,000 open-source machine learning models without managing GPU infrastructure. Built around Cog, its open-source model packaging tool, it covers image, text, audio, and video models on a pay-per-second billing model.

Audio

Stable Audio

Stable Audio generates royalty-safe instrumental music and sound effects from text prompts at 44.1kHz stereo quality. Built by Stability AI on a fully licensed dataset, it excels at background tracks, ambient beds, and sound design -- not vocal songs.

Video

LTX Studio

LTX Studio is Lightricks' AI filmmaking platform that turns scripts into storyboarded videos with consistent characters across scenes. It runs on LTX-Video, their open-source real-time video model, the fastest in its class on consumer hardware.

Video

Mochi 1

Mochi 1 is Genmo's 10 billion parameter open-source video generation model, released under Apache 2.0 in October 2024. It generates 480p clips at 30fps with strong prompt adherence. Weights are free to download and run commercially without royalties.

Image

Civitai

Civitai is the dominant community platform for sharing and discovering open-source AI image models. It hosts hundreds of thousands of Stable Diffusion and FLUX checkpoints, LoRAs, and embeddings, with an integrated cloud generator and in-browser LoRA training for artists without local GPU hardware.

Code

LlamaIndex

LlamaIndex is an open-source data framework for building RAG applications over private data, developed by LlamaIndex, Inc. It includes 300+ data connectors, built-in hierarchical chunking, and LlamaParse for enterprise-grade document parsing. Free OSS core; LlamaCloud starts at $50/mo.

AI Agents & Automation

LangChain

LangChain is the most widely adopted open-source framework for building LLM agents and RAG systems, with 135,000+ GitHub stars. Its October 2025 1.0 release brought stability guarantees, but persistent abstraction overhead and a paid LangSmith dependency make the calculus complicated.

AI Models & LLMs

Llama

Llama is Meta's family of open-weight language models, from the 2023 originals through Llama 4's multimodal MoE releases. Free to download and self-host, the models power thousands of derived tools and enterprise pipelines worldwide.

AI Models & LLMs

Mistral AI

Mistral is genuinely fast, notably cheaper than ChatGPT Pro, and the only frontier AI with Apache 2.0 open-weight models you can run entirely off-premises. But the quality gap on hard reasoning tasks is real, and the "European sovereign AI" story is more complicated than the marketing suggests.

AI Agents & Automation

Taiga

Taiga is an open-source agile project management platform built for multi-functional teams. It combines Scrum and Kanban boards with an AI assistant that automates task breakdown. While the desktop experience is highly customizable, the mobile application is just a web wrapper that lacks native functionality.

Image

GET3D by NVIDIA

GET3D by NVIDIA is a generative AI model that synthesizes high-quality 3D textured meshes from 2D image collections. It targets game developers and film studios needing rapid asset creation. It exports standard OBJ and GLB files. However, it requires at least 16GB of VRAM to run.

Audio

Suno AI Bark

Suno AI Bark is a transformer-based text-to-audio model for developers and creators. It generates speech, music, and non-verbal sounds across 13 languages. Users can add realistic laughs and sighs to audio. The 13-second context window restricts long-form narration.

Code

Sourcegraph

Sourcegraph is an AI code intelligence platform built for enterprise developers managing massive codebases. It searches across millions of repositories using regex and symbol support. Self-hosting requires significant server resources, making it less practical for small teams with limited infrastructure budgets.

Bonitasoft

Bonita is an open-source business process management platform that helps enterprise teams design and automate complex workflows. Developers use its visual drag-and-drop interface and BPMN 2.0 compliance to build custom applications. While the Community Edition offers unlimited users, the platform lacks the native AI capabilities found in newer competitors.

Ollama

Ollama is a local LLM runner that executes open large language models entirely on personal hardware. It offers zero API costs and full data privacy.

AI Agents & Automation

OpenClaw

OpenClaw is an open-source AI agents platform that automates web and workflow tasks. It runs locally or on a VPS without license fees, but token costs vary.

Audio

Amurex

Ditch premium subscriptions for meeting notes. This free open-source alternative records Google Meet in silence without annoying bots, but skips Zoom support.