Skip to main content

Overview

Venice API usage is pay per use. Text and decision requests are billed per token, on up to four rates:
  • Input: tokens you send to the model.
  • Output: tokens the model generates.
  • Cache read: input tokens served from the prompt cache, at a lower rate. See Prompt caching.
  • Cache write: input tokens written to the cache, on models that list this rate.
Embeddings are billed on input tokens only. Image, audio, music and video models are billed per unit: per image, per edit, per million characters, per second of audio, per track or per second of video.

How pricing works

  • Prices are in US dollars. Token prices are per 1 million tokens.
  • Some text models charge more for long prompts. Once a request’s input passes the threshold in Long-context rates, the higher rates apply to the whole request.
  • Web search, web scraping and X search on chat completions are billed on top of token prices. See Web search and tools.
  • Each model has its own price; there are no base or premium tiers. Some models are limited to paid plans (Pro, Plus and Max), and other accounts receive 401 with PRO_ONLY_MODEL.
  • Beta models can change or be removed without notice. Retiring models have a removal date; see Deprecations.
  • Prices come from model_spec.pricing in GET /models and, for video, from POST /video/quote. The models page shows the same prices, with filters to compare models.

Text models

Prices per 1M tokens. Privacy is the model’s privacy tier.

Long-context rates

These models switch to higher rates once a request’s input passes the threshold. The higher rates then apply to the whole request.

Decision models

Decision models answer typed questions about a state you send, instead of generating text. See Decisions. Prices per 1M tokens.

Embeddings

Embeddings are billed on input tokens. Prices per 1M tokens.

Images

Generation

Prices per image. Models with resolution options list a price for each resolution.

Generation price by quality

These models take a quality parameter. A request without one is billed at the default tier, the price in the generation table.

Editing

Prices per edit. The edit price covers the first input image. Models with an extra input image price charge it for each image after the first, so an edit with 3 input images at $0.11 per edit and $0.0035 per extra image costs $0.11 + 2 × $0.0035 = $0.117.

Editing price by quality

A request without a quality is billed at the default tier, the price in the editing table.

Upscaling

Audio

Text to speech

Billed per million characters of input text, on POST /audio/speech.

Speech to text

Billed per second of audio, on POST /audio/transcriptions.

Music, sound effects and long-form speech

These models run as queued jobs. Models priced by length charge one price per length tier. Get the exact price for a job from POST /audio/quote. The voice changer bills by the length of the source recording; quote it with POST /audio/voice-changer/quote.

Video

Prices per second of video. The rate depends on resolution and audio, so each row runs from the cheapest setting to the most expensive one. Get the exact price for a job from POST /video/quote before you queue it; each model’s page lists clip prices. Models priced by source charge by the video you send, so quote them with your input.

Web search and tools

Web search, web scraping and X search are billed on top of the model’s token price. Web scraping reads up to 5 URLs found in the user message and bills only the URLs it scrapes. X search bills each search result the model returns, so 10 results cost $0.10.

Paying for usage

  • API requests are pay per use at the rates on this page. There is no unmetered API tier, and a paid Venice plan does not include free or unlimited API usage. Free accounts can call the API at the same rates.
  • Pay with USD credits bought by card or crypto, which never expire; with staked DIEM, where each DIEM gives $1 of credit per day; or with USDC from a wallet through x402, with no account. See VVV & DIEM.
  • Venice charges each request to DIEM first, then earned credits, then bundled credits, then USD. Each balance is worth one dollar per unit.
  • When those balances cannot cover a request, the API returns 402 with INSUFFICIENT_BALANCE. See Error codes.
  • Paid plans include a monthly bundled-credit allowance: $1 on Pro, $75 on Plus and $225 on Max. Pro, Plus and Max subscribers also get a one-time $5 USD credit on the API balance, except accounts that upgraded by staking VVV.
  • Rate limits depend on the model and on your Default or Partner tier, not on the model’s price. See Rate limits.