Ollama

Ollama runs open models on your own machine. It needs no API key, and EdgeQuake talks to its native API. This page covers the setup, the Docker address and how to check the link.

When to use

• You want no API costs and no text leaving your machine.
• You accept slower extraction than a cloud model gives.
• You want a local default for development. Use OpenAI or Anthropic for higher quality.

Configure

ollama serve
ollama pull gemma4:latest
ollama pull embeddinggemma
export EDGEQUAKE_LLM_PROVIDER=ollama
export EDGEQUAKE_LLM_MODEL=gemma4:latest
export EDGEQUAKE_EMBEDDING_PROVIDER=ollama
export OLLAMA_EMBEDDING_MODEL=embeddinggemma
export OLLAMA_HOST=http://127.0.0.1:11434

Variable	Purpose	Default
OLLAMA_HOST	Server URL	http://localhost:11434
EDGEQUAKE_LLM_MODEL	Chat model. Pull it first.	none
OLLAMA_EMBEDDING_MODEL	Embedding model for EDGEQUAKE_EMBEDDING_PROVIDER=ollama	embeddinggemma:latest

Models

• Chat: any model you have pulled. The catalog lists gemma4:latest, gemma3:latest and llama3.2:latest, among others.
• Embeddings: embeddinggemma is the quickstart default and returns 768-dimension vectors. nomic-embed-text is also in the catalog.

Changing the embedding model changes the vector size. Rebuild stored embeddings after you change it (see Roles).

Docker

A container cannot reach 127.0.0.1 on your host. Use http://host.docker.internal:11434 instead. quickstart.sh rewrites loopback addresses for you and prints the address it chose.

Verify

curl http://127.0.0.1:11434/api/tags
curl http://127.0.0.1:8080/health

/health calls /api/version on Ollama. The probe (POST /api/v1/providers/test with "shape":"ollama") checks /api/version and sends a short chat request to /api/chat. It skips the embedding check.

%%{init: {"theme":"base","themeVariables":{"primaryColor":"#E0E7FF","primaryBorderColor":"#6366F1","primaryTextColor":"#1E1B4B","secondaryColor":"#D1FAE5","secondaryBorderColor":"#10B981","secondaryTextColor":"#064E3B","tertiaryColor":"#FEF3C7","tertiaryBorderColor":"#F59E0B","tertiaryTextColor":"#6B7A90","lineColor":"#7A889C","clusterBkg":"rgba(99,102,241,0.07)","clusterBorder":"#A5B4FC","noteBkgColor":"#FEF9C3","noteTextColor":"#422006","textColor":"#6B7A90","titleColor":"#6B7A90","signalColor":"#7A889C","signalTextColor":"#6B7A90","loopTextColor":"#6B7A90","edgeLabelBackground":"#F1F5F9","actorLineColor":"#94A3B8"}}}%%
%% eq-theme:v1
sequenceDiagram
  participant U as Operator
  participant EQ as EdgeQuake
  participant OL as Ollama
  U->>EQ: GET /health
  EQ->>OL: GET /api/version
  alt Ollama answers
    OL-->>EQ: version
    EQ-->>U: llm_provider true
  else No answer
    EQ-->>U: llm_provider false, status degraded
  end

The health check reports degraded when Ollama does not answer. Each request is a live call, not a stored status.

Troubleshoot

• Extraction fails with a network error. Ollama is not running. Start it with ollama serve and check Troubleshooting.
• Probe reports shape_mismatch (“chat ping failed”) for a model you pulled. The chat model name is wrong or not pulled. /api/version does not list models, so the probe cannot report model_not_found for Ollama. Run ollama list and compare the name.
• A saved Connection points at another Ollama host. The Connection is tested and used with its own URL. The environment default and /health still read OLLAMA_HOST. Set both when they must agree.
• Ingestion is slow. EdgeQuake limits extraction concurrency for local providers such as Ollama. This is expected on a laptop.

Related: Providers overview, Anthropic.