Skip to main content
The Model API is OpenAI-compatible. If you already use the OpenAI Python or Node.js SDK, you can switch to Jarvis by changing the base_url and API key — no other code changes needed. All requests require an Authorization header. See the API overview for authentication details.

​
Model naming

Jarvis runs models via Ollama, and model names follow the provider/model convention:
Model nameDescription
ollama/llama3Meta Llama 3 (8B or 70B depending on your node)
ollama/mistralMistral 7B — fast, general-purpose
ollama/deepseek-coderDeepSeek Coder — optimized for code tasks
ollama/phi3Microsoft Phi-3 — efficient, low-memory
ollama/gemma2Google Gemma 2
Use GET /models to see the exact list available on your instance.

​
GET /models

List all models currently available for inference.

​
Response

​
object
string
Always "list".
​
data
array
An array of model objects.

​
Example

curl
curl https://your-jarvis-host/models \
  -H "Authorization: Bearer jrv_yourkey123"
Response
{
  "object": "list",
  "data": [
    {
      "id": "ollama/llama3",
      "object": "model",
      "created": 1712000000,
      "owned_by": "ollama"
    },
    {
      "id": "ollama/mistral",
      "object": "model",
      "created": 1712000000,
      "owned_by": "ollama"
    }
  ]
}

​
POST /chat/completions

Send a multi-turn conversation to a model and receive a completion. This endpoint is fully OpenAI-compatible.

​
Request

​
model
string
required
The model to use (e.g., ollama/llama3). Use GET /models to see available options.
​
messages
array
required
An array of message objects representing the conversation history.
​
stream
boolean
If true, the response streams as server-sent events. Defaults to false.
​
temperature
number
Sampling temperature between 0 and 2. Higher values produce more varied output. Defaults to 1.
​
max_tokens
integer
Maximum number of tokens to generate. Defaults to model-dependent limits.

​
Response

​
id
string
Unique identifier for this completion.
​
object
string
Always "chat.completion".
​
model
string
The model that generated the response.
​
choices
array
An array of completion choices (usually one).
​
usage
object
Token usage for the request.

​
Examples

curl -X POST https://your-jarvis-host/chat/completions \
  -H "Authorization: Bearer jrv_yourkey123" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ollama/llama3",
    "messages": [
      { "role": "system", "content": "You are a helpful infrastructure assistant." },
      { "role": "user", "content": "What are common causes of high disk I/O on Linux?" }
    ],
    "temperature": 0.7
  }'
Response
{
  "id": "chatcmpl-9v2kxp3z",
  "object": "chat.completion",
  "model": "ollama/llama3",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Common causes of high disk I/O include..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 34,
    "completion_tokens": 112,
    "total_tokens": 146
  }
}

​
POST /completions

Generate a text completion from a raw prompt string (non-chat format).

​
Request

​
model
string
required
The model to use (e.g., ollama/mistral).
​
prompt
string
required
The input text to complete.
​
max_tokens
integer
Maximum tokens to generate.
​
temperature
number
Sampling temperature. Defaults to 1.
​
stream
boolean
Stream the response as server-sent events. Defaults to false.

​
Response

​
id
string
Unique identifier for this completion.
​
object
string
Always "text_completion".
​
choices
array
An array of completion choices.

​
Example

curl
curl -X POST https://your-jarvis-host/completions \
  -H "Authorization: Bearer jrv_yourkey123" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "ollama/mistral",
    "prompt": "Write a one-paragraph summary of the benefits of local AI inference:",
    "max_tokens": 150
  }'
For most tasks, prefer POST /chat/completions. The chat format gives the model more context about the conversation role and produces better results with instruction-tuned models.