Token usage How much the model is used. Position in OpenRouter's rolling seven-day token ranking. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Requests How often the model is called. Position in OpenRouter's rolling seven-day request ranking. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Tokens per request How heavy a single call to this model is. The tokens one call to this model moved, prompt and completion together, averaged over the week. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Request trend Whether people call this model more this week than last week. How many more, or fewer, calls this model took this week than last week. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Intelligence How capable the model is. The model's own score on the Epoch Capabilities Index. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Intelligence / $ How much capability your money buys. Epoch Capabilities Index points divided by blended price per million tokens. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Price What the model costs to run. The cheapest eligible offer, blended 75% input and 25% output, in USD per million tokens. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Providers How many places you can buy it. How many providers list a public pay-as-you-go price for the model. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Context How much text the model can hold at once. tokens The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Parameters How big the model is. The total parameters the model's creator published. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Output speed How fast the model writes its answer, on the route it is priced from. The median output tokens per second on the route the price is taken from, over that route's last half hour. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset
Released How long ago the model arrived. The days since the model was released, counted on the day this publication was built. The bright part of a bar counts the models your other filters keep. Narrowing this row hides models that have no value here. That is missing data, not a low score. Reset