Price Comparison API

GET/v1/prices

Public endpoint - no authentication required. Returns a price comparison matrix across all providers for every model in the marketplace.

The endpoint takes no parameters and always returns the full matrix. Filter client-side on models[].model.

Examples

curl https://api.surplusintelligence.ai/v1/prices

curl $SURPLUS_BASE_URL/prices

{
  "models": [
    {
      "model": "grok-4.5",
      "displayName": "Grok 4.5",
      "modelProvider": "xAI",
      "providers": [
        {
          "provider": "bankr",
          "providerName": "Bankr",
          "providerModelId": "grok-4.5",
          "pricing": { "input": 2, "output": 6, "cacheRead": 0.5, "webSearch": 0.005 }
        },
        {
          "provider": "openrouter",
          "providerName": "OpenRouter",
          "providerModelId": "x-ai/grok-4.5",
          "pricing": { "input": 2, "output": 6, "cacheRead": 0.3, "webSearch": 0.005 }
        },
        {
          "provider": "venice",
          "providerName": "Venice AI",
          "providerModelId": "grok-4-5",
          "pricing": { "input": 2.27, "output": 6.8, "cacheRead": 0.34 }
        }
      ]
    }
  ],
  "providers": ["anthropic", "bankr", "bedrock", "deepseek", "engy", "fireworks", "google", "groq", "jatevo", "mistral", "mordiem", "morpheus", "openai", "openrouter", "together", "uncensored", "venice", "zai"],
  "updated_at": "2026-09-15T13:00:00.000Z"
}

Fields

Model Entry

modelstring
Standard marketplace model name
displayNamestring
Human-readable name
modelProviderstring
Model creator (Anthropic, OpenAI, xAI, Google, etc.)
providersarray
Reference-pricing entries for catalog-mapped providers with publishable pricing for this model (see Provider Entry)
mediaTypestring
Media class for media-metered models (optional; text models omit it)
mediaPricingobject
Per-unit media reference pricing (optional; present with mediaType; see Media Pricing)

Provider Entry

providerstring
Provider key (e.g. openrouter)
providerNamestring
Human-readable provider name
providerModelIdstring
The model id as that provider names it
pricingobject
That provider's price entry (see below)

Price Entry (`providers[].pricing`)

input, output, cacheRead, cacheWrite, and reasoning are in USD per 1M tokens. Other optional fields state their own unit below.

inputnumber
Input/prompt token price
outputnumber
Output/completion token price
cacheReadnumber
Cache hit price (optional)
cacheWritenumber
Cache creation price (optional)
reasoningnumber
Thinking/reasoning token price (optional)
webSearchnumber
Web-search price (optional) - USD per web-search request. Present only for providers that publish web-search pricing (currently openrouter, bedrock, and bankr); omitted elsewhere.
imagenumber
Image/media price (optional). For vision text models without mediaUnit, this is USD per 1M image tokens; when mediaUnit is present, this is USD per that unit.
imageInputPerMpnumber
Image-input price in USD per megapixel (optional)
imageOutputPerMpnumber
Generated-image price in USD per output megapixel (optional)
mediaUnitstring
Human-readable unit for image media pricing (optional), such as image, 1K characters, minute, 1M tokens, job, or megapixel

providers[] is ordered by pricing.output, ascending - the cheapest output price first. That ordering is a rough convenience, not a total cost ranking; compare entries yourself if you need a specific workload's cost.

Media Pricing (`mediaPricing`)

Media model rows use a separate mixed-unit pricing object. All fields are present as numbers; a value of 0 means that pricing dimension does not apply to the model.

perImageUsdnumber
USD per generated image
per1kCharsUsdnumber
USD per 1,000 characters
perMinuteUsdnumber
USD per minute of media
per1MTokensUsdnumber
USD per 1M tokens
inputPer1MUsdnumber
USD per 1M input tokens
outputPer1MUsdnumber
USD per 1M output tokens
outputPerMpUsdnumber
USD per output megapixel

One model, many provider prices

provider-acheapest output$0.29 /1M out

provider-b$0.34 /1M out

provider-cbut may be cheaper on input$0.58 /1M out

The array is sorted by output price as a convenience. It is not a total-cost ranking - your actual cost depends on your input/output mix, so sort it yourself if that matters.

Caching

The response is cached for 5 minutes (Cache-Control: public, max-age=300). The updated_at field indicates when the response was generated - at most ~5 minutes before you received it.

Reference-Pricing Providers

The live providers array is the sorted set of provider keys that appear in at least one providers[] pricing row. A row is published only when the catalog maps the model to a recognized provider and a positive reference price is available.

This is reference-pricing coverage, not seller onboarding or marketplace liquidity. A listed provider may have no active seller offer, and an onboarded seller/provider may be absent. Do not use this array to infer current availability, offer count, or provider health; treat the live response as the source of truth for the current reference-pricing set.

Interactive Price Table

See the live price comparison API for the current machine-readable matrix. Treat that response as the source of truth for current provider rows and counts.