API Reference: Model Management

List, show, create, copy, delete, pull, push models.

List Models — GET /api/tags

curl http://localhost:11434/api/tags
curl https://ollama.com/api/tags  # Cloud models (requires OLLAMA_API_KEY)

Response: ListResponse

{
  "models": [
    {
      "name": "qwen3:1.7b",
      "model": "qwen3:1.7b",
      "remote_model": "",
      "remote_host": "",
      "modified_at": "2025-10-03T23:34:03Z",
      "size": 1400000000,
      "digest": "sha256:a2af6cc3eb7f...",
      "details": {
        "format": "gguf",
        "family": "qwen3",
        "families": ["qwen3"],
        "parameter_size": "1.7B",
        "quantization_level": "Q4_K_M"
      }
    }
  ]
}

ModelSummary Fields

FieldTypeDescription namestringModel identifier modelstringModel identifier (same as name) remote_modelstringUpstream model name (if remote/cloud) remote_hoststringUpstream Ollama host URL (if remote) modified_atstringISO 8601 last modified sizeintegerDisk size in bytes digeststringSHA256 digest details.formatstringFile format ("gguf") details.familystringPrimary model family details.familiesstring[]All applicable families details.parameter_sizestringe.g., "1.7B", "7B" details.quantization_levelstringe.g., "Q4_K_M", "Q8_0"

---

Show Model Details — POST /api/show

curl http://localhost:11434/api/show -d '{"model": "qwen3:1.7b"}'

Verbose (includes full model_info)
curl http://localhost:11434/api/show -d '{"model": "qwen3:1.7b", "verbose": true}'

Request

ParameterTypeRequiredDescription modelstringyesModel name verbosebooleannoInclude full metadata

Response: ShowResponse

FieldTypeDescription parametersstringModel parameter settings as text licensestringLicense text modified_atstringISO 8601 timestamp detailsobjectformat, family, families, parameter_size, quantization_level templatestringPrompt template (Go template syntax) capabilitiesstring[]["completion", "vision", "tools", "thinking"] model_infoobjectDetailed architecture metadata

Example Response

{
  "parameters": "temperature 0.7\nnum_ctx 2048",
  "capabilities": ["completion", "vision"],
  "details": {
    "format": "gguf",
    "family": "gemma3",
    "parameter_size": "4.3B",
    "quantization_level": "Q4_K_M"
  },
  "model_info": {
    "general.architecture": "gemma3",
    "general.parameter_count": 4299915632,
    "gemma3.context_length": 131072,
    "gemma3.embedding_length": 2560,
    "gemma3.block_count": 34
  }
}

Key for mindX: The capabilities array tells you exactly what a model supports — use this for dynamic capability detection instead of hardcoded lists.

---

Create Model — POST /api/create

Create a custom model from an existing one with modified system prompt, parameters, etc.

curl http://localhost:11434/api/create -d '{
  "model": "mindx-agent",
  "from": "qwen3:1.7b",
  "system": "You are mindX, an autonomous multi-agent orchestration system.",
  "parameters": {"temperature": 0.7, "num_ctx": 4096}
}'

Request

ParameterTypeRequiredDescription modelstringyesName for the new model fromstringnoBase model to derive from systemstringnoSystem prompt templatestringnoPrompt template licensestring\string[]noLicense text parametersobjectnoDefault parameter overrides messagesChatMessage[]noConversation examples quantizestringnoQuantization level: q4_K_M, q8_0, etc. streambooleannoStream status updates

Quantize an Existing Model

curl http://localhost:11434/api/create -d '{
  "model": "qwen3:1.7b-q8",
  "from": "qwen3:1.7b-instruct-fp16",
  "quantize": "q8_0"
}'

---

Copy Model — POST /api/copy

curl http://localhost:11434/api/copy -d '{
  "source": "qwen3:1.7b",
  "destination": "qwen3-backup"
}'

Useful for creating OpenAI-compatible aliases:

curl http://localhost:11434/api/copy -d '{
  "source": "qwen3:1.7b",
  "destination": "gpt-3.5-turbo"
}'

---

Delete Model — DELETE /api/delete

curl -X DELETE http://localhost:11434/api/delete -d '{"model": "old-model"}'

---

Pull Model — POST /api/pull

Download from Ollama registry.

# Streaming (default) — shows progress
curl http://localhost:11434/api/pull -d '{"model": "qwen3:1.7b"}'

Non-streaming
curl http://localhost:11434/api/pull -d '{"model": "qwen3:1.7b", "stream": false}'

Status Events (streaming)

{"status": "pulling manifest"}
{"status": "downloading sha256:abc...", "digest": "sha256:abc...", "total": 1400000000, "completed": 700000000}
{"status": "verifying sha256 digest"}
{"status": "writing manifest"}
{"status": "success"}

---

Push Model — POST /api/push

Publish to Ollama registry (requires authentication).

curl http://localhost:11434/api/push -d '{"model": "username/my-model"}'

---

Blobs

Check Blob Exists — HEAD /api/blobs/:digest

curl -I http://localhost:11434/api/blobs/sha256:29fdb92e57cf...
200 = exists, 404 = not found

Upload Blob — POST /api/blobs/:digest

curl -T model.gguf -X POST http://localhost:11434/api/blobs/sha256:29fdb92e57cf...
201 = created, 400 = digest mismatch

---

mindX Integration

# Via OllamaHandler (llm/ollama_handler.py)
handler = OllamaHandler(model_name_for_api="qwen3:1.7b")

List models
models = await handler.list_local_models_api()

Get model info
info = await handler.get_model_info_api("qwen3:1.7b")

Pull a model (with progress)
await handler.pull_model_api("deepseek-r1:1.5b")

Via OllamaChatManager — automatic discovery
await chat_manager.discover_models(force=True)
available = chat_manager.available_models

---