DSPy
Self-Hosted AIFramework for programming, not prompting, language models
Replaces LangChain
The 12 best No-Database self-hosted ai you can self-host, ranked by community traction.
Framework for programming, not prompting, language models
Replaces LangChain
Graph-based retrieval-augmented generation system
Replaces OpenAI Assistants
AI developer that writes apps with human oversight
Replaces Devin
Chat with your documents fully offline
Replaces ChatGPT
Constrained generation language for controlling LLMs
Replaces OpenAI API
Language-model agent that resolves GitHub issues
Replaces Devin
Unit testing framework for LLM outputs
Replaces LangSmith, Braintrust
Evaluation framework for retrieval-augmented generation
Replaces LangSmith
Structured text generation for language models
Replaces OpenAI Structured Outputs
Embedded vector database for AI applications
Replaces Pinecone
High-quality multilingual text-to-speech library
Replaces ElevenLabs, Amazon Polly
State-of-the-art music source separation
Replaces LALAL.AI
No apps match these filters.
| App | Deploy effort | Difficulty | Min. RAM | License | ★ Stars |
|---|---|---|---|---|---|
| DSPy | Read-the-docs project | Medium | 1,024 MB | MIT | 38,620 |
| GraphRAG | Read-the-docs project | Hard | 4,096 MB | MIT | 36,327 |
| GPT Pilot | Read-the-docs project | Medium | 1,024 MB | MIT | 33,652 |
| LocalGPT | Under-an-hour setup | Medium | 8,192 MB | Apache-2.0 | 22,192 |
| Guidance | Read-the-docs project | Medium | 8,192 MB | MIT | 21,797 |
| SWE-agent | Under-an-hour setup | Medium | 2,048 MB | MIT | 20,546 |
| DeepEval | Read-the-docs project | Medium | 1,024 MB | Apache-2.0 | 18,841 |
| Ragas | Read-the-docs project | Medium | 1,024 MB | Apache-2.0 | 16,015 |
| Outlines | Under-an-hour setup | Medium | 8,192 MB | Apache-2.0 | 15,927 |
| LanceDB | ≈5-minute setup | Easy | 512 MB | Apache-2.0 | 11,666 |
Every option here is open-source and self-hostable, tagged No-Database within the self-hosted ai category. Deploy effort is graded from published container recipes, not vendor claims — see the methodology.