Openrouter Model Catalog

Key fields:

• pricing.prompt / pricing.completion -- cost per token (not per million; multiply by 1M for readable rates)
• context_length -- max input tokens
• top_provider.max_completion_tokens -- max output tokens
• architecture.modality -- text->text, text+image->text, etc.

Python: Query and Filter

import requests

models = requests.get("https://openrouter.ai/api/v1/models").json()["data"]

# Find all free models
free_models = [m for m in models if m["pricing"]["prompt"] == "0"]
print(f"Free models: {len(free_models)}")

# Models with tool calling support
# (query with supported_parameters)
tool_models = requests.get(
    "https://openrouter.ai/api/v1/models?supported_parameters=tools"
).json()["data"]
print(f"Tool-calling models: {len(tool_models)}")

# Sort by prompt price (cheapest first, excluding free)
paid = [m for m in models if float(m["pricing"]["prompt"]) > 0]
paid.sort(key=lambda m: float(m["pricing"]["prompt"]))
for m in paid[:10]:
    cost_per_m = float(m["pricing"]["prompt"]) * 1_000_000
    print(f"  ${cost_per_m:.2f}/M tokens — {m['id']} ({m['context_length']//1000}K ctx)")

# Filter by context length (128K+)
large_ctx = [m for m in models if m["context_length"] >= 128_000]
print(f"128K+ context models: {len(large_ctx)}")

List Providers for a Model

# See all providers and their pricing for a specific model
curl -s "https://openrouter.ai/api/v1/models/anthropic/claude-3.5-sonnet/endpoints" | jq '.data[] | {
  provider: .provider_name,
  price_prompt: .pricing.prompt,
  price_completion: .pricing.completion,
  context_length: .context_length,
  quantization: .quantization
}'

Model Variants

Append a suffix to any model ID for variant behavior:

Special Routers

# Let OpenRouter pick the best model
response = client.chat.completions.create(
    model="openrouter/auto",
    messages=[{"role": "user", "content": "Write a SQL query to find duplicate emails"}],
    max_tokens=200,
)
print(f"Auto-selected: {response.model}")  # Shows which model was chosen

Popular Model Quick Reference

Prices change frequently. Always verify via /api/v1/models.

Error Handling

Enterprise Considerations

• Cache the model catalog with 1-hour TTL (model availability changes infrequently)
• Build a model allowlist for your organization to restrict which models teams can use
• Monitor /api/v1/models for deprecation notices and new model additions
• Use supported_parameters query filter to ensure models support features you need (tools, JSON mode, etc.)
• Compare providers via the endpoints API to find the cheapest or fastest provider for each model

References

• Examples | Errors
• Models Docs | Models API | Model Variants