Pulse score (60d)
0–100 composite · dailyCollecting history — comes back as pulse rolls up daily.
Local LLM runtime with an OpenAI-compatible API. Agents swap OpenAI for an Ollama endpoint to run on-prem or on a laptop.
Ollama (MIT) packages llama.cpp with a daemon that serves one or more open-weight models over HTTP. One container, one `/api/chat` endpoint, OpenAI-compatible drop-in on `/v1/chat/completions`.
Who uses it: solo developers who want local inference without cloud bills, teams that need on-prem for compliance reasons, agent authors building self-hostable products ("bring your own LLM"), and anyone running a vision/code model on a single beefy workstation. GPU acceleration (CUDA, Metal, ROCm) is auto-detected. Alternatives: vLLM (higher throughput, more ops), llama.cpp direct, LocalAI (feature-wider but heavier).
Collecting history — comes back as pulse rolls up daily.
Collecting history — pull snapshots started today; the chart populates over the next few days.
No comments yet. Be the first.