Models#

Oru-el provides access to 100+ leading open-source and commercial models. All models are available through the same API — just change the model parameter.

Listing models#

Via the API#

Fetch the full model catalog (no authentication required):

curl https://api.oru-el.com/v1/inference/models

Response:

{
  "models": [
    {
      "id": "llama-4-maverick",
      "displayName": "Llama 4 Maverick",
      "description": "Meta's latest open-source model...",
      "modelType": "TEXT_GENERATION",
      "category": "CHAT",
      "creator": "Meta",
      "parameterCount": "17B",
      "contextWindow": 131072,
      "quantization": null,
      "isFeatured": true,
      "capabilities": ["chat", "tool_calling", "json_mode"],
      "pricing": {
        "standard": {
          "inputPerMToken": "0.10",
          "outputPerMToken": "0.30",
          "cachedInputPerMToken": "0.05"
        },
        "turbo": {
          "inputPerMToken": "0.20",
          "outputPerMToken": "0.60",
          "cachedInputPerMToken": null
        }
      }
    }
  ]
}

Get a specific model#

curl https://api.oru-el.com/v1/inference/models/llama-4-maverick

This returns additional details including benchmarks, architecture info, and links.

In the dashboard#

Browse the Model Catalog in the dashboard or visit the public catalog at oru-el.com/catalog to explore models with a visual interface. Each model card shows capabilities, pricing, context window, and a link to try it in the Playground.

Model fields#

Model types#

Model categories#

Categories describe what a model is optimized for:

Choosing the right model#

By task#

By priority#

Cost vs. quality tradeoff#

Larger models produce higher-quality output but cost more per token. For many tasks, smaller models are sufficient:

• 1B-3B models — Classification, simple extraction, routing
• 7B-8B models — General chat, summarization, translation
• 27B-70B models — Complex reasoning, nuanced writing, code generation
• 70B+ models — State-of-the-art quality for the hardest tasks

Pricing tiers#

Each model has one or two pricing tiers:

Standard tier#

The default tier. Requests are routed for the lowest cost.

Turbo tier#

Low-latency optimized routing for faster responses at a higher price. Enable with:

response = client.chat.completions.create(
    model="llama-4-maverick",
    messages=[{"role": "user", "content": "Hello!"}],
    extra_body={"tier": "turbo"},
)

Not all models have a turbo tier. Check the pricing.turbo field in the model response — if it's null, only standard routing is available.

Context windows#

The context window determines the maximum number of tokens (input + output) a model can handle in a single request. Context windows vary by model:

If your input exceeds the context window, the API will return an error. Plan your message history and system prompts accordingly.

Using a model in code#

Every model uses the same API — just swap the model parameter:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.oru-el.com/v1/inference",
    api_key="oruel_your_api_key_here",
)

# Use any model by its ID
response = client.chat.completions.create(
    model="llama-4-maverick",
    messages=[{"role": "user", "content": "Hello!"}],
)

# Switch to a different model — no other code changes needed
response = client.chat.completions.create(
    model="deepseek-r1-0528",
    messages=[{"role": "user", "content": "Solve this step by step: what is 15% of 340?"}],
)

Featured models#

Featured models are highlighted in the catalog for their quality, popularity, and reliability. They're a good starting point if you're not sure which model to choose. Look for models with isFeatured: true in the API response.