Kimi K3: input -33%, output -23%. New rate: 0.30 / 11.59 dollars per million tokens
The OpenRouter list price for Kimi K3 has gone down. Input now costs 0.30 dollars per million tokens (previously 0.45) and output 11.59 dollars (previously 15.00). For a typical request with 1,500 input and 2,200 output tokens that is a change of 23 percent. OpenRouter routes each model to several hosting providers, so a change like this can come from a different host rather than from the model maker.
👁 117 reads
The OpenRouter list price for Kimi K3 has gone down. Input now costs 0.30 dollars per million tokens (previously 0.45) and output 11.59 dollars (previously 15.00). For a typical request with 1,500 input and 2,200 output tokens that is a change of 23 percent. OpenRouter routes each model to several hosting providers, so a change like this can come from a different host rather than from the model maker.
In practice: generating a 1500-word article on this model costs about 0.0259 USD. For comparison, the cheapest models I currently track are Qwen 3.7 Flash (0,13 USD per million output tokens) and DeepSeek V4 Flash (0,64 USD).
A complete comparison of rates for all tracked models, converted to real tasks and USD, is updated daily in the AI cost calculator. Rates come from the OpenRouter public API and may differ from promotions at the provider itself.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →More on this from QUANTSLANT
ai tools·Nemotron 3.5 Super VL vs Step 5: Weights, Licenses and TestTwo October releases with different deployment needs and license terms. Nemotron completed all 12 Polish tasks; Step returned three empty answers.

