Replicate
Open models as one-line APIs — image, video and voice, billed by the second.
Why it's listed
When you want to test open models without deployment hassle, Replicate gives you one-line APIs for image, video, and voice. Per-second billing keeps prototype costs low and saves you from infrastructure work.
In the same category
OpenRouter
One API key, hundreds of models — compare, fall back and rate-limit without the plumbing.
Ollama
The lowest bar for local open models — one command for Llama, Qwen or DeepSeek; data stays on your machine.
Hugging Face
The GitHub of models and datasets — find models, check leaderboards, run demos.
Firecrawl
Clean web-to-markdown/structured-data for LLMs — essential plumbing for content pipelines and RAG.
Related
Linked by topic, people and hubs
Local or cloud: record choice conditions without a benchmark
ViewDify knowledge-Q&A tutorial: supported inquiry drafts and unknowns
ViewOllama Quickstart: run an open model locally
ViewTransformers Can't Think In Physics: Anima Anandkumar on Accelerated Understanding
ViewFound a great tool?
Submit it — after human review it joins the AI Directory.
Submit a tool