Simulator
What your inference will cost
Build a month of requests from any mix of models: chat and reasoning, embeddings, images, video, speech and transcription. See the total and how it grows, where each model ranks against every other model of its kind, and what the same model costs on other hosts.
How the numbers are worked out
Each line is one kind of request, repeated a number of times a month. The cost of a request uses the same price book and rules as the Veqen dashboard: cached input at the cache-read rate, batch rates where the provider offers them, images by size and quality, video by the second, speech by the character and transcription by the second of audio.
The ladder under each line places that request among every current model of the same kind, priced for exactly the same usage. Same model on other hosts matches open models such as Llama, Qwen, DeepSeek and gpt-oss across Together, Fireworks, Groq, DeepInfra, Bedrock and the rest. A cheaper model isn't always an equivalent one, so test quality before you switch.
List prices aren't always what you pay. In the dashboard, the same simulator starts from your own usage and applies your negotiated discounts and self-hosted prices.
Already spending? Run a free audit of what you actually pay.