ODOCK.AI
Models & MCPModels

Model Pricing

Understand model pricing fields, pricing types, and how Odock calculates model usage cost.

Model Pricing

Model pricing is used to calculate cost in usage records and to enforce budgets. Pricing is stored on the model so billing follows the model that was actually used.

In the UI, model pricing is edited as human-readable USD per 1M tokens. Internally, the gateway normalizes usage from the upstream provider, then calculates cost using the model pricing snapshot.

Pricing can include input text cost, output text cost, image cost, audio cost, video cost, reasoning cost, cached input discounts, embeddings cost, fine-tuning cost, tool-related cost metadata, and notes.

Pricing Types

Pricing typeWhat it representsUsed when
Input textCost for prompt, message, or text input tokens.Chat, reasoning, completion, and text-processing requests.
Output textCost for generated text output tokens.Chat, reasoning, completion, and text generation responses.
Input imageCost for image input tokens or image input units.Vision requests where the model receives images.
Output imageCost for generated image output tokens or image output units.Image generation, image edits, or image variation endpoints.
Input audioCost for audio input tokens or units.Audio understanding, transcription, or multimodal requests with audio input.
Output audioCost for generated audio output tokens or units.Text-to-speech or multimodal responses with audio output.
Input videoCost for video input tokens or units.Multimodal models that accept video input.
Output videoCost for generated video output tokens or units.Video generation or multimodal responses with video output.
ReasoningCost for reasoning or thinking tokens reported by the provider.Reasoning models that expose a separate reasoning token count.
Cached input discountPercentage discount applied to cached input tokens.Providers that report cache hits or cached prompt tokens.
EmbeddingsCost for embedding tokens.Embedding endpoints where usage is counted separately from text generation.
Fine-tuning trainingCost for tokens used while training or fine-tuning a model.Fine-tuning workflows when training cost is tracked.
Fine-tuning usageCost for tokens used by a fine-tuned model.Runtime calls to fine-tuned models with separate rates.
Tool pricingOptional cost metadata for tool definitions, tool inputs, or tool outputs.Model calls that include tools and where tool-related token costs need separate reporting.
NotesHuman-readable pricing notes.Explaining provider-specific assumptions, contract pricing, or known limitations.

Usage Calculation

Usage records keep normalized tokens, computed billing, and the raw provider usage payload for reconciliation.

For usage monitoring, see Usage Monitoring. For budgets, see Budgets. To change pricing in the UI, see Edit model pricing.

On this page