Pricing Guide

OpenModels shows model pricing so developers can compare cost before routing traffic. Token-priced routes list input and output token prices. Other routes can use request, image, or endpoint-specific pricing units.

How to read model prices

The Models page shows these columns:

Token-priced values use /1M. Request-priced and image-priced routes show /req or /image in the Models table.

Provider routes

Each model can have one or more provider routes. If you omit route, OpenModels uses the default available route for the selected model. If you send route.provider, usage is billed against that selected provider route.

Use provider names you have confirmed in the Models table or OpenModels dashboard:

{
  "model": "qwen3.5-flash",
  "route": { "provider": "<provider-for-this-model>" },
  "route_mode": "balanced",
  "messages": [{ "role": "user", "content": "Hello!" }]
}

Cost formula

For token-priced chat models, request cost is based on actual token usage:

cost = (input_tokens / 1,000,000 * input_price)
     + (output_tokens / 1,000,000 * output_price)

Example:

input_tokens = 10,000
output_tokens = 2,000
input_price = $0.029 / 1M tokens
output_price = $0.295 / 1M tokens
cost = 10,000 / 1,000,000 * 0.029
     + 2,000 / 1,000,000 * 0.295

The final credit deduction uses the actual usage reported for the request.

Input vs output price

Output tokens are often more expensive than input tokens. A model that looks cheap for input-heavy workloads may cost more for long-generation workloads.

When comparing models, consider:

• Expected prompt size
• Expected output length
• Whether the model needs reasoning
• Whether the model needs tool calling or vision
• Context window requirements
• Reliability and route availability

Context is not price

The context window is the maximum amount of text the model can consider. A larger context window does not mean every request costs more by itself. Cost depends on the tokens actually sent and generated.

Tiered pricing

Some models use tiered pricing. In that case, the applicable price can depend on usage thresholds or model-specific pricing rules. Check the Models table and route details before using tiered models in production.

Choosing a model by cost

For development and high-volume classification, start with a lower-cost model such as qwen3.5-flash when it fits the task.

For complex reasoning, tool use, long context, or vision, choose a model whose capabilities match the task even if the token price is higher. A cheaper model is not cheaper if it requires multiple retries or produces unusable results.

Price changes

Model availability and pricing can change as supply, route availability, and provider pricing change. Check the Models page before launching a new production workload.