Tiledesk
74Open-source live chat and chatbot platform for customer service, run in their cloud or self-hosted
Best for: Teams required to keep chat data on their own servers
ai-for / self-hosted
As of , EffectHub tracks 32 AI tools for self hosted; 29 (91%) have a free tier, and paid plans run from $4 to $150 a month.
Tiledesk, AsoFast and Tines lead the list on click-through. 25 of the 32 ship a public API.
Self-hosted AI tools run on your own servers or cloud account instead of the vendor's, so data never leaves infrastructure you control. The trade is operational: you provide the compute (often a GPU for local models), apply updates and handle backups. For teams with confidentiality obligations or data-residency rules that trade is worth it; for everyone else, compare the real cost of running it against the hosted plan.
ranked by click-through
Open-source live chat and chatbot platform for customer service, run in their cloud or self-hosted
Best for: Teams required to keep chat data on their own servers
Self-hosted, open-source tool that generates app store listing copy, translations, and screenshots using your own AI API key.
Best for: Indie developers already comfortable self-hosting small services
Workflow automation and AI agents for security and IT teams, with a free tier capped at three live workflows
Best for: Security operations teams automating triage, enrichment and response
Open-source chat UI that lets you plug your own API keys into ChatGPT, Replicate, and image models from one interface.
Best for: developers who already hold API keys for OpenAI or Replicate
Open-source agent platform now sold as a team of AI specialists coordinated in chat, with a visual workflow builder underneath.
Best for: Teams that want to self-host an agent platform and own the infrastructure
ByteDance's MIT-licensed agent framework: sandboxed agents that research, code and edit files.
Best for: Teams building long-running research or coding agents they want to self-host
Free, self-hosted face-swap and enhancement software, released under a responsible-use AI license.
Best for: Developers and researchers who want a local, controllable face-swap pipeline
Self-hosted bookmark manager that auto-tags saved links, notes, and images, and makes them full-text and semantically searchable.
Best for: People who already self-host other services and want their bookmarks in the same stack
Open-source workspace for building AI agents from a visual canvas, a prompt, or code
Best for: Teams that want a visual agent builder they can also self-host
Open-source workspace for docs, wikis and databases, with AI chat and a self-hosted option.
Best for: Teams that need their notes and project data on their own infrastructure
Self-hosted plain-English search across your own videos and photos, with face grouping, indexed and queried entirely offline.
Best for: People with large local video and photo archives who refuse to upload them
Open-source, self-hosted AI assistant you message from WhatsApp, Telegram, Slack and 10+ other chat apps
Best for: Developers and tinkerers who want a hackable assistant with system access
Self-hostable enterprise agent platform for document RAG, database queries and voice AI, with data-sovereignty controls.
Best for: Regulated organizations such as finance, healthcare or government-adjacent teams that need enterprise chatbots on infrastructure they control
Open-source AI coding agent and autocomplete that runs in your IDE, in the cloud or fully self-hosted
Best for: Teams whose code must stay on-prem but still want an agent and completions
Open-weight image models, free to run below $1M revenue
Best for: Developers and studios that need generation on their own infrastructure with no per-call cost
Workflow automation with AI-agent nodes that runs on n8n's cloud or self-hosted from GitHub
Best for: Technical teams that want to self-host automation with Docker, on-prem or air-gapped
MIT-licensed workflow and AI-agent automation with 760+ integrations, on its EU cloud or your own servers
Best for: Teams that want Zapier-style flows without sending data to a vendor's cloud
A self-hosted proxy that checks role, redacts fields and logs every request an AI agent makes against databases, files and APIs.
Best for: Platform and security teams putting agents in front of production databases
Spreadsheet-style workbench for writing, testing and shipping prompts, with a firewall on the outputs
Best for: Teams where product or support staff own prompt wording, not just engineers
Tracing, analytics and prompt versioning for LLM apps, with a self-hostable community edition.
Best for: Small teams debugging agent traces and watching token spend
Company-wide agent platform reachable from Slack, email and iMessage, with adaptive model routing and an isolated computer per run.
Best for: Engineering orgs that want one agent reachable from Slack, GitHub, Linear and email
Open-source canvas for building LLM workflows, agents and RAG pipelines, hosted or self-run
Best for: Teams that want to self-host an agent and RAG builder without writing the orchestration layer
Open-source machine learning plus an enterprise agentic AI platform that runs on your own hardware
Best for: Data science teams that want a free, distributed AutoML and classical ML library with R and Python bindings
Self-hosted open-source runtime that serves text, speech, vision, image and agent workloads behind OpenAI-compatible APIs.
Best for: Teams that must keep inference on their own infrastructure
Self-hosted AI assistant with persistent memory that ingests email, files, and calendars, free to run yourself.
Best for: Privacy-conscious users willing to self-host with Docker
Prebuilt GPU-optimized model containers with standard APIs for self-hosting AI models, plus hosted endpoints for prototyping.
Best for: Enterprises self-hosting LLMs and other models on their own NVIDIA GPUs
Open-source AI chat and search layer that connects any LLM to a company's apps, with permission-aware retrieval and self-hosting.
Best for: IT teams that must keep documents and embeddings inside their own network
Self-hosted email marketing platform with an AI writer, list management, and deliverability tools like content spinning.
Best for: Marketers who want email sent from infrastructure they control
Personal AI assistant with persistent memory that acts across email, calendar, files and chat apps
Best for: Individuals who want one assistant with memory across email, calendar and chat
Open-source Python SDK and self-hosted runtime for building agents, teams and workflows, with a web control plane to monitor them.
Best for: Python developers building product copilots, research or support agents they will host themselves
Open-source tracing and eval layer that turns production agent failures into signals your coding agent can fix and verify.
Best for: Teams running agents in production who want traces, detectors and evals in one open-source tool
Open-source CRM you can self-host, with custom objects, a no-code workflow builder and a native MCP server
Best for: Technical teams that want to own their CRM data and schema
| Tool | Price | Free plan | API | Status |
|---|---|---|---|---|
| Tiledesk | Not confirmed | Yes | Yes | Active |
| AsoFast | Free | Yes | Not confirmed | Active |
| Tines | Not confirmed | Yes | Yes | Active |
| Anse | Free | Yes | Not confirmed | Active |
| AutoGPT | from $42.50/mo | Yes | Yes | Active |
| DeerFlow | Free | Yes | Yes | Unstable |
| FaceFusion | from $20 once | Yes | Not confirmed | Active |
| Karakeep | from $4/mo | Yes | Yes | Active |
| Sim | from $25/seat/mo | Yes | Yes | Active |
| AppFlowy | from $12/seat/mo | Yes | Yes | Active |
| Hachi | Not confirmed | Yes | Not confirmed | Active |
| OpenClaw | Free | Yes | Not confirmed | Active |
| Raghim AI | On request | Not confirmed | Yes | Active |
| Refact.ai | from $10/mo | Yes | Not confirmed | Active |
| Stable Diffusion | Not confirmed | Yes | Yes | Active |
| n8n | Not confirmed | Yes | Yes | Active |
| Activepieces | from $20/mo | Yes | Yes | Active |
| InterLock by Context Data | On request | Not confirmed | Yes | Active |
| Langtail | from $99/mo | Yes | Yes | Active |
| Lunary | from $20/seat/mo | Yes | Yes | Active |
| Adapt | from $14/seat/mo | Yes | Yes | Active |
| Dify | from $59/mo | Yes | Yes | Active |
| H2O.ai | Not confirmed | Yes | Yes | Active |
| LocalAI | Free | Yes | Yes | Active |
| Mantle | Free | Yes | Yes | Active |
| NVIDIA NIM | Not confirmed | Yes | Yes | Active |
| Onyx | from $20/seat/mo | Yes | Yes | Active |
| Sendster | from $27/mo | No | Not confirmed | Active |
| Vellum | from $30/mo | Yes | Yes | Active |
| Agno | from $150/mo | Yes | Yes | Active |
| TraceRoot | Not confirmed | Yes | Yes | Active |
| Twenty | from $9/seat/mo | Yes | Yes | Active |
Yes. 6 of the 32 AI tools for self hosted on EffectHub's list are free to use — AsoFast, Anse, DeerFlow and OpenClaw; 23 more have a free tier with paid upgrades.
As of October 27, 2026, paid plans for the 32 AI tools for self hosted on EffectHub's list run from $4 to $150 a month. Each card shows the pricing model and the cheapest paid plan as EffectHub last verified it; 29 of the 32 can be started for free.
25 of the 32 AI tools for self hosted on EffectHub's list ship a public API, including Tiledesk, Tines, AutoGPT and DeerFlow. The API badge on each card is re-checked against the vendor's documentation.
Control over data and cost: documents, prompts and outputs stay inside your infrastructure, which satisfies confidentiality and residency requirements, and at scale running your own instance can be cheaper than per-seat cloud pricing. The cost is operating it yourself.
Only if they run models locally. Many self-hosted tools call a hosted model API and need only an ordinary server; tools that run open models on your hardware need a GPU with enough memory for the model you choose.
For journalists, researchers and AI assistants: the figures on this page in one sentence you can quote.
As of October 27, 2026, EffectHub tracks 32 AI tools for self hosted; 29 (91%) have a free tier, and paid plans run from $4 to $150 a month. (EffectHub, “Self-Hosted AI Tools”, Oct 2026, https://effecthub.net/ai-for/self-hosted)
Figures as of How we count
These figures are free to use under CC BY 4.0 with a link to this page.